WO2021249461A1 - 制冷设备控制方法、装置、计算机设备和计算机可读介质 - Google Patents

制冷设备控制方法、装置、计算机设备和计算机可读介质 Download PDF

Info

Publication number
WO2021249461A1
WO2021249461A1 PCT/CN2021/099313 CN2021099313W WO2021249461A1 WO 2021249461 A1 WO2021249461 A1 WO 2021249461A1 CN 2021099313 W CN2021099313 W CN 2021099313W WO 2021249461 A1 WO2021249461 A1 WO 2021249461A1
Authority
WO
WIPO (PCT)
Prior art keywords
refrigeration equipment
air conditioner
neural network
sample data
preset
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/CN2021/099313
Other languages
English (en)
French (fr)
Inventor
刘明明
熊勇
胡先红
林东华
秦世好
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
ZTE Corp
Original Assignee
ZTE Corp
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by ZTE Corp filed Critical ZTE Corp
Priority to JP2022576079A priority Critical patent/JP7473690B2/ja
Priority to EP21821753.7A priority patent/EP4166862A4/en
Priority to BR112022025218A priority patent/BR112022025218A2/pt
Publication of WO2021249461A1 publication Critical patent/WO2021249461A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F24—HEATING; RANGES; VENTILATING
    • F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
    • F24F11/00—Control or safety arrangements
    • F24F11/62—Control or safety arrangements characterised by the type of control or by internal processing, e.g. using fuzzy logic, adaptive control or estimation of values
    • F24F11/63—Electronic processing
    • F24F11/64—Electronic processing using pre-stored data
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F24—HEATING; RANGES; VENTILATING
    • F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
    • F24F11/00—Control or safety arrangements
    • F24F11/30—Control or safety arrangements for purposes related to the operation of the system, e.g. for safety or monitoring
    • F24F11/32—Responding to malfunctions or emergencies
    • F24F11/38—Failure diagnosis
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F24—HEATING; RANGES; VENTILATING
    • F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
    • F24F11/00—Control or safety arrangements
    • F24F11/30—Control or safety arrangements for purposes related to the operation of the system, e.g. for safety or monitoring
    • F24F11/46—Improving electric energy efficiency or saving
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F24—HEATING; RANGES; VENTILATING
    • F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
    • F24F11/00—Control or safety arrangements
    • F24F11/70—Control systems characterised by their outputs; Constructional details thereof
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F24—HEATING; RANGES; VENTILATING
    • F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
    • F24F11/00—Control or safety arrangements
    • F24F11/50—Control or safety arrangements characterised by user interfaces or communication
    • F24F11/61—Control or safety arrangements characterised by user interfaces or communication using timers
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F24—HEATING; RANGES; VENTILATING
    • F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
    • F24F2110/00—Control inputs relating to air properties
    • F24F2110/10—Temperature
    • F24F2110/12—Temperature of the outside air
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F24—HEATING; RANGES; VENTILATING
    • F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
    • F24F2130/00—Control inputs relating to environmental factors not covered by group F24F2110/00
    • F24F2130/10—Weather information or forecasts
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F24—HEATING; RANGES; VENTILATING
    • F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
    • F24F2140/00—Control inputs relating to system states
    • F24F2140/50—Load

Definitions

  • the present disclosure relates to the field of automatic control technology, and in particular to a refrigeration equipment control method, device, computer equipment and computer readable medium.
  • the present disclosure provides a refrigeration equipment control method, which includes: determining the current outdoor temperature; and inputting historical sample data of the refrigeration equipment load and preset influence factors as the first input parameters into a first neural network model to obtain the forecast of the refrigeration equipment on the day Load; the historical sample data of the outdoor temperature and the load predicted on the day of the refrigeration equipment as the second input parameters are input into the second neural network to obtain the predicted indoor temperature of the day; the predicted indoor temperature of the day and the preset cooling
  • the efficiency factor is input into the third neural network as the third input parameter to obtain the optimal control parameter of the refrigeration equipment of the day; and the operation of the refrigeration equipment is controlled according to the optimal control parameter.
  • the present disclosure also provides a refrigeration equipment control device, including: a first processing module, a second processing module, and a control module, the first processing module is used to determine the current outdoor temperature; the second processing module is used to: The historical sample data of the equipment load and the preset influencing factors are input into the first neural network model as the first input parameters to obtain the load predicted on the day of the refrigeration equipment; the historical sample data of the outdoor temperature and the load predicted on the day of the refrigeration equipment are combined Input the second neural network as the second input parameter to obtain the predicted indoor temperature of the day; input the predicted indoor temperature of the day and the preset cooling efficiency factor as the third input parameter into the third neural network to obtain the refrigeration equipment on the day
  • the optimal control parameter the control module is used to control the operation of the refrigeration equipment according to the optimal control parameter.
  • the present disclosure also provides a computer device including: one or more processors; and a storage device on which one or more programs are stored; when the one or more programs are executed by the one or more processors At this time, the one or more processors are caused to implement the refrigeration device control method according to the present disclosure.
  • the present disclosure also provides a computer-readable medium on which a computer program is stored, wherein, when the program is executed by a processor, the processor enables the processor to implement the refrigeration device control method according to the present disclosure.
  • Figure 1 is a schematic diagram of a refrigeration equipment control system provided by the present disclosure
  • Figures 2 to 4 are schematic diagrams of the procedures for establishing the first, second, and third neural network models provided by the present disclosure
  • FIG. 5 is a schematic diagram of the control flow of the refrigeration equipment provided by the present disclosure.
  • FIGS. 6A to 6C are schematic diagrams of the first, second, and third neural network models provided by the present disclosure.
  • FIG. 7 is a schematic diagram of the air-conditioning control process provided by the present disclosure.
  • Figure 8 is a schematic diagram of the control flow of the heat exchange equipment provided by the present disclosure.
  • Fig. 9 is a schematic diagram of the process of re-determining and updating the optimal control parameters of the refrigeration equipment on the day provided by the present disclosure.
  • FIG 10 and 11 are schematic diagrams of the structure of the refrigeration equipment control device provided by the present disclosure.
  • the conventional linkage control algorithm is a traditional temperature control start-stop method.
  • the ambient temperature is used as the main basis for the linkage control of heat exchange equipment and air conditioners.
  • the algorithm is simple but difficult to improve.
  • the conventional linkage control process is as follows: real-time detection of indoor and outdoor temperature, if the indoor temperature exceeds the upper limit of the equipment operation temperature, start the heat exchange equipment or air conditioning refrigeration: when the heat exchange equipment startup conditions are met (such as the indoor and outdoor temperature difference reaches the threshold), priority is started Heat exchange equipment, otherwise start the air conditioner. Air conditioning and heat exchange equipment should not be switched frequently, with an interval of more than half an hour.
  • the start and stop condition parameters of the heat exchange equipment and the air conditioner separately.
  • the start and stop temperature of the heat exchange equipment can be 35/25°C, and the temperature difference is 8°C.
  • the temperature difference between indoor and outdoor exceeds 8°C, it is allowed to start the heat exchange equipment.
  • the start-stop condition parameters are not fixed. If a fixed start-stop condition parameter is set, it will cause Frequent activation of air conditioners increases energy consumption.
  • the conventional linkage control algorithm only considers the external factor of the ambient temperature. The start time and number of starts of the air conditioner are unpredictable, and the control accuracy is low, so it is very difficult to improve.
  • the air conditioner needs to be turned on, but if the high temperature time above 40°C can be predicted in advance, it will be very short and will not affect the safe operation of the equipment (the working range of some base stations/transmission equipment can reach 40°C for a long time, and 50°C for a short time) In fact, there is no need to turn on the air conditioner. In this way, while ensuring the safety of the equipment, it is possible to avoid turning on the air conditioner once, and to achieve a certain degree of energy saving.
  • the present disclosure provides a method for controlling refrigeration equipment, which can control the operation of refrigeration equipment in a computer room.
  • the method can be applied to the refrigeration control system shown in FIG. 1.
  • the refrigeration control system provided by the present disclosure includes a refrigeration equipment control device, a field supervision unit (FSU), and a refrigeration equipment.
  • FSU is a field device, set in the computer room where the refrigeration equipment is located, and includes a collection unit and an execution unit.
  • the collection unit is used to collect real-time data such as outdoor temperature and humidity, indoor temperature, equipment load, etc., and upload it to the refrigeration control device.
  • the execution unit is used to control the operation of the refrigeration equipment according to the instruction of the refrigeration control device.
  • the refrigeration equipment control device can be a cloud device, and a Unified Management Expert (UME) can be selected, which is configured with a first neural network (NN) model, a second neural network model, and a third neural network model , Historical sample database and control strategy of refrigeration equipment (for example, refrigeration control algorithm).
  • UME can obtain the predictive control plan of the refrigeration equipment according to the data reported by the FSU and the first neural network model, the second neural network model, and the third neural network model, and issue the predictive control plan to the FSU.
  • the refrigeration equipment can include air conditioning and heat exchange equipment, and can operate according to the issued control plan.
  • the following first threshold to tenth threshold and various durations can be preset in the refrigeration equipment control device.
  • the first threshold VHT may be 45°C.
  • the second threshold VLT for example, may be 15° C., when the indoor temperature is lower than VLT, the air conditioner is unconditionally turned off, and the second threshold VLT is less than the first threshold VHT.
  • the third threshold HT AC may be, for example, 40°C.
  • the fourth threshold HT HEE for example, may be 35°C.
  • the fifth threshold is used to determine whether the second high temperature pre-start condition of the indirect heat exchange equipment is met.
  • the sixth threshold LT may be 25°C.
  • the sixth threshold LT is less than the fourth threshold HT HEE and the third threshold HT AC .
  • the seventh threshold is used to determine how long the refrigeration equipment is down.
  • the eighth threshold is used to determine whether the indoor and outdoor temperature difference in the second high temperature pre-start condition of the direct heat exchange equipment is met.
  • the ninth threshold is used to determine whether the humidity in the second high temperature pre-start condition of the direct heat exchange equipment is met.
  • the tenth threshold is used to judge the error between the actual operating parameters of the air conditioner and the optimal control parameters of the air conditioner.
  • the maximum air conditioner on time MAXCOT and the air conditioner's shortest off time MINCST is Generally, the air conditioner's maximum on time MAXCOT is 12 hours, and the air conditioner's shortest off time MINCST is 0.5 hours.
  • the first neural network model, the second neural network model and the third neural network model are established.
  • the flow of establishing the first neural network model, the second neural network model, and the third neural network model will be described in detail below in conjunction with FIG. 2.
  • the establishment of the first neural network model, the second neural network model, and the third neural network model includes the following steps S21 to S23.
  • step S21 historical sample data is acquired.
  • Sample data can include outdoor temperature, indoor temperature, and refrigeration equipment load.
  • the refrigeration equipment control device may obtain historical sample data from the historical database.
  • the historical database can store a large amount of historical sample data such as daily outdoor temperature T Rout , indoor temperature T Rin , refrigeration equipment load L R and so on. Can vary the degree of urgency determine the sampling period, e.g., the outdoor temperature T Rout sampling period may be 10 minutes, the sampling period T Rin room temperature and refrigeration load L R five minutes may be based on these parameters.
  • the sample data may include analog data and sampling data.
  • the simulated data is data obtained by simulating the operation of the refrigeration equipment when the indoor temperature is greater than the third threshold HT AC.
  • the sampled data is the data sampled when the indoor temperature is less than the sixth threshold LT and the actual shutdown duration of the refrigeration equipment is greater than the seventh threshold. That is to say, when the indoor temperature T Rin is high and the refrigeration equipment needs to be operated, the dummy load can be used to simulate the real refrigeration equipment, and data such as T Rout , T Rin , L R and so on can be recorded.
  • T Rin is low and the refrigeration equipment is out of service for a long time (such as the season or night when the outdoor temperature T Rout is low)
  • a large amount of existing historical sample data can be directly used to speed up the collection of historical sample data.
  • step S22 the historical sample data is simulated and simulated, and the daily optimal control parameters of the refrigeration equipment are calculated.
  • step S22 through computer simulation training, establish the heat distribution map of the computer room environment, heating equipment and refrigeration equipment, simulate and calculate the historical sample data, and output the optimal solution vector of the refrigeration equipment control of the day (that is, the daily optimal refrigeration equipment Control parameters), and save the daily optimal control parameters of the refrigeration equipment as sample data tags.
  • the air conditioner should not be turned on frequently.
  • the air conditioner can be turned on at most 12 times a day, and the heat exchange equipment may be turned on at most 12 times a day.
  • T moment /T hours T moment /T hours
  • the optimal control parameter of the air conditioner for that day is: the air conditioner is turned on and run twice every day. T moment is turned on at the turn-on time, and the running time is the value of the corresponding turn-on time T hours .
  • step S23 a first neural network model, a second neural network model, and a third neural network model are established based on historical sample data and daily optimal control parameters of the refrigeration equipment.
  • step S23 the first neural network model, the second neural network model and the third neural network model are established in sequence.
  • the refrigeration equipment control method of the present disclosure may further include steps S22' to S23'.
  • step S22' the historical sample data and the daily optimal control parameters of the refrigeration equipment are normalized.
  • the historical sample data and the daily optimal control parameters of the refrigeration equipment can be normalized according to the following formula, so that the data is in the range (0, 1):
  • X real is the true value of the actual sample
  • X * is the normalized data
  • X max is the maximum or upper limit of the corresponding type of data sample
  • X min is the minimum or lower limit of the corresponding type of data sample value.
  • X max is the upper limit may be 100 °C
  • X min may be a lower limit -40 °C
  • a training sample data set is established based on the normalized data.
  • the training sample data set includes a training set, a verification set, and a test set.
  • step S23' a training set, a verification set, and a test set can be established according to a sample ratio of 6:2:2.
  • step S23 the step of establishing the first neural network model, the second neural network model, and the third neural network model according to the historical sample data and the daily optimal control parameters of the refrigeration equipment (ie, step S23) may include: according to the training sample The data set establishes the first neural network model, the second neural network model and the third neural network model.
  • the steps of establishing the first neural network model, the second neural network model, and the third neural network model may include steps S231 to S233.
  • step S231 the historical sample data of the load of the refrigeration equipment and the preset influence factor are used as the first input parameters, and the historical sample data of the load of the refrigeration equipment of the day is used as the first output parameter to establish a first neural network model.
  • the impact factor can include one or any combination of the following: holiday impact factor F holiday , tide impact factor F tide , and regional event factor F event .
  • holiday impact factor F holiday The value ranges of holiday influencing factor F holiday , tide influencing factor F tide and regional event factor F event are all (0, 1), and can be determined based on manual experience.
  • the holiday impact factor F holiday on normal working days can be 0, the holiday impact factor F holiday on weekends can be 0.1, and the holiday impact factor F holiday on the Spring Festival holiday can be 0.25, etc.; for industrial parks , The tidal impact factor F tide during working hours can be 0.5, the tide impact factor F tide during overtime can be 0.7, and the tide impact factor F tide during the late night period can be 0.3, etc.; for some areas, normal areas
  • the event factor F event can be 0, the regional event factor F event for commercial marketing activities can be 0.1, the regional event factor F event for gatherings can be 0.2, the regional event factor F event for concerts can be 0.3, and so on.
  • step S232 the historical sample data of the same period of outdoor temperature and the historical sample data of the load on the day of the refrigeration equipment are used as the second input parameter, and the historical sample data of the indoor temperature of the day is used as the second output parameter to establish a second neural network model.
  • step S233 the historical sample data of the indoor temperature of the day and the preset cooling efficiency factor are used as the third input parameter, and the historical sample data of the optimal control parameter of the refrigeration equipment of the day is used as the third output parameter to establish a third neural network model .
  • the optimal control parameters may include the opening time T moment and the opening time T hours , that is, the air conditioner opening time T moment-AC , the heat exchange device opening time T moment-TEE , the air conditioner opening time T hours-AC and the heat exchange device opening time T hours hours-TEE .
  • the refrigeration efficiency factor may include a heat exchange refrigeration efficiency factor F eff1 and an air conditioning refrigeration efficiency factor F eff2 .
  • F eff1 and F eff2 are both constant. If the environment of the computer room changes (for example, the cooling equipment is replaced or the space position is moved, etc.), the heat exchange and cooling efficiency factor must be changed F eff1 and the air-conditioning refrigeration efficiency factor F eff2 are adjusted to new constants.
  • T moment1 is 0.45
  • T hours1 is 0.05
  • T moment2 is 0.60
  • T hours2 is 0.10
  • the first neural network model, the second neural network model, and the third neural network model are trained and optimized, they can be deployed according to the actual operating environment.
  • the three neural network models can all be deployed on UME to make full use of the powerful computing resources of the cloud to achieve real-time or online training. If necessary, the three neural network models can also be deployed on the edge side by adding computing sticks, for example, on the FSU.
  • Fig. 5 is a schematic diagram of the control flow of the refrigeration equipment provided by the present disclosure.
  • the refrigeration equipment control method provided by the present disclosure can be used to control the operation of the refrigeration equipment, and includes steps S11 to S15.
  • step S11 the current outdoor temperature is determined.
  • the current outdoor temperature T Rout can be calculated by weighting according to the predicted temperature and the detected outdoor temperature, that is, the outdoor temperature within the preset time period before the current time is first determined, and then the outdoor temperature within the preset time period before the current time and the predicted temperature of the day are determined. And the preset first weight and second weight to determine the current outdoor temperature T Rout .
  • the preset duration may be 1 hour
  • FSU can collect indoor and outdoor temperature, humidity, refrigeration equipment load and other data and upload it to UME.
  • step S12 the historical sample data of the load of the refrigeration equipment and the preset influence factors are input as the first input parameters into the first neural network model to obtain the load predicted on the day of the refrigeration equipment.
  • the same period refers to the same period in history. For example, the same moment of today last year and the same moment of today the year before can be the same period of today.
  • step S12 as shown in FIG. 6A, input the historical sample data L N of the load of the refrigeration equipment, the holiday influencing factor F holiday , the tide influencing factor F tide and the regional event factor F event into the first neural network model to obtain the refrigeration equipment
  • the load L R predicted on the day is used as the output value of the first neural network model.
  • step S13 the historical sample data of the outdoor temperature and the load predicted on the day of the refrigeration equipment are input as the second input parameters into the second neural network to obtain the predicted indoor temperature on the day.
  • step S13 as shown in FIG. 6B, input the outdoor temperature historical sample data T Rout and the load L R (ie, the output value of the first neural network) predicted on the day of the refrigeration equipment into the second neural network model to obtain The predicted indoor temperature T Rin on the day is used as the output value of the second neural network model.
  • step S14 the predicted indoor temperature of the day and the preset refrigeration efficiency factor are input into the third neural network as the third input parameters to obtain the optimal control parameters of the refrigeration equipment on the day.
  • step S14 as shown in FIG. 6C, the predicted indoor temperature T Rin (that is, the output value of the second neural network), the heat exchange and refrigeration efficiency factor F eff1, and the air conditioning and refrigeration efficiency factor F eff2 of the day are input to the third neural network.
  • the model obtains the optimal control parameters of the air conditioner on the day (ie, the air conditioner on time T moment-AC and the air conditioner on time T hours-AC ) and the optimal control parameters of the heat exchange equipment on the day (ie, the heat exchange device on time T moment- TEE and heat exchange equipment open time T hours-TEE ).
  • the opening time T moment can be rotated to the hh:mm:ss format, and the opening time T hours can be rotated to the standard time length (for example, xx hours).
  • the UME runs the first neural network model, the second neural network model, and the third neural network model in sequence to output the optimal control parameters of the refrigeration equipment of the day.
  • the optimal control parameters of the air conditioner may include the air conditioner on time T moment-AC and the air conditioner on time T hours-AC
  • the optimal control parameters of the heat exchange equipment may include the heat exchange equipment on time T moment-TEE and the heat exchange equipment on time T hours-TEE .
  • the optimal control parameters can include up to 12 sets of air conditioner on time T moment-AC and air conditioner on time T hours-AC every day , and up to 24 sets of heat exchange equipment on time T moment-TEE and heat exchange equipment on time T hours-TEE .
  • step S15 the operation of the refrigeration equipment is controlled according to the optimal control parameters.
  • step S15 the operation of the air conditioner is controlled according to the optimal control parameter of the air conditioner, and the operation of the heat exchange device is controlled according to the optimal control parameter of the heat exchange device.
  • the neural network model is used to combine current outdoor temperature, refrigeration equipment load historical sample data, influence factors and refrigeration efficiency factors and other parameters to realize the prediction and linkage of control schemes for air conditioning and heat exchange equipment Control
  • the predicted control scheme has high accuracy, overcomes the defects of traditional algorithms that are difficult to improve, realizes active control of air conditioning and heat exchange equipment, optimizes operating efficiency, and reduces energy consumption
  • the chronological data is compared with the current actual measurement
  • the combination of data and taking into account the influencing factors of special events and the influencing factors of the refrigeration efficiency of refrigeration equipment makes the predicted control scheme more accurate, adaptable to changes in the environment of the computer room, and enhance the scope of application.
  • Fig. 7 is a schematic diagram of the air-conditioning control process provided by the present disclosure.
  • the air conditioning control process provided by the present disclosure includes steps S31 to S39.
  • step S31 if the current indoor temperature is greater than the first threshold VHT, step S36 is executed; otherwise, step S32 is executed.
  • step S31 if the current indoor temperature is greater than VHT, indicating that the current indoor temperature is too high, it can be further judged whether the air conditioner is running overtime (ie, step S36); if the current indoor temperature is less than or equal to VHT, it can be further judged Whether the temperature is too low (ie, execute step S32).
  • step S32 if the current indoor temperature is less than the second threshold VLT, step S39 is executed; otherwise, step S33 is executed.
  • step S32 if the current indoor temperature is less than the second threshold VLT, indicating that the current indoor temperature is too low, the air conditioner can be shut down due to the abnormal low temperature (ie, step S39); if the current indoor temperature is greater than or equal to the second threshold VLT, It means that the current indoor temperature will not be shut down due to abnormal high temperature and will not shut down due to abnormal low temperature, and it can be further determined whether the first high temperature pre-start condition is met (ie, step S33 is executed).
  • step S33 if the first high temperature pre-start condition is met, step S34 is executed; otherwise, step S31 is returned.
  • step S33 the current indoor temperature is less than or equal to the first threshold VHT and greater than or equal to the second threshold VLT, and if the first high temperature pre-start condition is met, the air conditioner is controlled to operate according to the optimal control parameters of the day (ie, execute step S34); if the first high temperature pre-start condition is not met, return to step S31.
  • the first high-temperature pre-start condition may include: reaching the air conditioner on time T moment-AC , the current indoor temperature is greater than the third threshold HT AC , and the actual air conditioner shutdown duration is greater than the air conditioner's shortest shutdown duration MINCST.
  • step S34 the maximum air conditioner operating time T on-max is set to the minimum value of the air conditioner on time T hours-AC and the air conditioner maximum on time MAXCOT.
  • step S34 the minimum value of the air conditioner on time T hours-AC and the air conditioner’s maximum on time MAXCOT may be taken as the control parameter for actually controlling the operation of the air conditioner, so as to ensure the reliability and safety of the air conditioner operation.
  • Step S35 start the air conditioner, and execute step S38.
  • step S35 after the air conditioner is controlled to start, start to record the actual on-time duration of the air conditioner T on-AC , reset the actual off-time duration of the air conditioner T off-AC to zero, and execute step S38.
  • step S36 if the air conditioner's actual shutdown time T off-AC is greater than the air conditioner's shortest shutdown time MINCST, step S37 is executed; otherwise, step S31 is returned.
  • step S36 the current indoor temperature is greater than the first threshold VHT, if the current air conditioner actual shutdown duration T off-AC is greater than the air conditioner minimum shutdown duration MINCST, indicating that the high temperature abnormal start condition is met, then the air conditioner high temperature abnormal start operation is performed (ie, execute step 37); if the current actual air conditioner shutdown time T off-AC is less than or equal to the air conditioner's shortest shutdown time MINCST, return to step S31.
  • step S37 the maximum air conditioner operating time Ton-max is set to the maximum air conditioner on time MAXCOT, and step S35 is executed.
  • step S37 when the air conditioner is started abnormally at a high temperature, the operating time of the air conditioner is directly controlled according to the preset maximum on time MAXCOT of the air conditioner.
  • step S38 if the actual on-time of the air conditioner Ton-AC is greater than or equal to the maximum operating time of the air conditioner Ton-max , step S39 is executed; otherwise, the current state of the air conditioner is maintained.
  • the air conditioner After the air conditioner is started, it starts to record the actual air conditioner on time Ton-AC . If the air conditioner’s actual on time Ton-AC is greater than or equal to the maximum air conditioner operating time Toon-max , the air conditioner is turned off; otherwise, the current state of the air conditioner is maintained.
  • step S39 the air conditioner is turned off, and the process returns to step S31.
  • step S39 after controlling the air conditioner to turn off, start recording the actual off time T off-AC of the air conditioner, and clear the actual on time T on-AC of the air conditioner to zero, and then return to step S31 to continue detecting the indoor temperature.
  • the air-conditioning control process may further include: if the current indoor temperature is less than the sixth threshold LT, turning off the air-conditioning.
  • the air conditioner When the actual room temperature exceeds the first threshold VHT, the air conditioner can be started due to abnormal high temperature; when the actual room temperature is lower than the second threshold VLT, the air conditioner can be shut down due to the abnormal low temperature; when the air conditioner is turned on T moment-AC and the actual room temperature exceeds
  • the third threshold HT AC meets the requirement that the operation interval exceeds the shortest shutdown duration MINCST, the air conditioner will operate according to the prediction scheme output by the third neural network model, that is, the air conditioner will start to operate when T moment-AC arrives at the time when the air conditioner is turned on, and the operation duration is the air conditioner. Turn on time T hours-AC .
  • Figure 8 is a schematic diagram of the control flow of the heat exchange equipment provided by the present disclosure.
  • the heat exchange equipment control process provided by the present disclosure includes steps S41 to S44.
  • step S41 if the second high temperature pre-start condition is met, step S42 is executed; otherwise, the current state of the heat exchange equipment is maintained.
  • the heat exchange equipment may include direct heat exchange equipment and indirect heat exchange equipment
  • the direct heat exchange equipment may include a fresh air system
  • the indirect heat exchange equipment may include heat pipe equipment (HPE).
  • the second high-temperature pre-start condition may include: reaching the turning-on time T moment-TEE of the heat exchange equipment, and the current indoor temperature is greater than the fourth threshold HT HEE , and the current indoor temperature and outdoor temperature The difference is greater than the fifth threshold.
  • the second high temperature pre-start condition includes one of the following:
  • the eighth threshold may be greater than the fifth threshold, that is, in the second high-temperature pre-start condition, the indoor and outdoor temperature difference requirements of the direct heat exchange equipment are higher than the indoor and outdoor temperature difference requirements of the indirect heat exchange equipment.
  • the fifth threshold may be 6°C
  • the eighth threshold can be 10°C.
  • the second high temperature pre-start condition of the direct heat exchange device may include temperature conditions and humidity conditions, for example, the ninth threshold may be 90%.
  • step S42 the heat exchange equipment is started.
  • step S42 after the heat exchange equipment is controlled to start, it starts to record the actual turn-on time T on-HEE of the heat exchange equipment, and clears the actual shutdown time T off-HEE of the heat exchange equipment to zero.
  • step S43 if the actual opening time of the heat exchange equipment Ton-HEE is greater than or equal to the opening time T hours-HEE of the heat exchange equipment, step S44 is executed; otherwise, the current state of the heat exchange equipment is maintained.
  • step S44 the heat exchange equipment is turned off.
  • step S44 after controlling the heat exchange equipment to turn off, start recording the actual shutdown time T off-HEE of the heat exchange equipment, and clear the actual turn-on time T on-HEE of the heat exchange equipment to zero.
  • the control process of the heat exchange device may further include: if the current indoor temperature is less than the sixth threshold LT, turning off the heat exchange device.
  • the air conditioner and the indirect heat exchange equipment can operate at the same time, but the operation of the air conditioner and the direct heat exchange equipment are mutually exclusive, that is, the air conditioner and the direct heat exchange equipment cannot operate at the same time.
  • the air conditioner and the direct heat exchange equipment cannot operate at the same time.
  • the refrigeration device control method of the present disclosure may further include: if the air conditioner is turned on, the heat exchange device is turned off; if the heat exchange is turned on Equipment, turn off the air conditioner.
  • the air conditioning and heat exchange equipment control algorithm can be run on the UME cloud. If necessary, the air conditioning and heat exchange equipment control algorithm can also be copied to the FSU for local execution. In this case, the UME must first The refrigeration control plan predicted by the three neural network is issued to the FSU in advance.
  • Steps S11 to S14 shown in FIG. 5 are executed once a day before zero o'clock to output the optimal control parameters of the refrigeration equipment on the day.
  • FIG. 9 is a schematic diagram of the process of re-determining and updating the optimal control parameters of the refrigeration equipment on the day provided by the present disclosure.
  • the refrigeration equipment control method of the present disclosure may further include steps S51 to S53.
  • step S51 if the error between the actual operating parameters of the air conditioner on that day and the optimal control parameters of the air conditioner on that day exceeds the tenth threshold, step S52 is executed; otherwise, the process ends.
  • step S52 the optimal control parameters of the air conditioner for the day are determined again.
  • step S52 is the same as that of steps 12 to S14, and will not be repeated here.
  • step S53 the training sample data set is updated according to the re-determined optimal control parameters of the air conditioner on the day.
  • the parameter updates the training sample data set to improve the timely adaptability of the refrigeration control strategy and the real-time performance and accuracy of the predictive control.
  • Neural network models can be deployed and run in the cloud. When external parameters are constantly changing, these models can also be continuously trained in real time or online to continuously improve prediction accuracy, and can be trained and adjusted to adapt to abnormal conditions such as changes in the computer room environment.
  • the refrigeration equipment control method of the present disclosure may further include: if one type of refrigeration equipment currently running fails and the other type of refrigeration equipment is normal, turning off the failed refrigeration equipment and turning it on The normal refrigeration equipment; and if both of the two currently running refrigeration equipment are faulty, when the fault is eliminated, the refrigeration equipment with the elimination of the fault is activated. That is to say, if the currently opened refrigeration equipment fails, the faulty refrigeration equipment is turned off and the normal refrigeration equipment is turned on. When the fault is eliminated, the refrigeration equipment is turned on and the other refrigeration equipment is turned off. Through the mutual backup startup operation of the air conditioner and the heat exchanger in the event of failure, the danger of abnormally high temperature in the computer room can be avoided.
  • the refrigeration equipment control method of the present disclosure may further include: if the current indoor temperature is less than the sixth threshold LT and the refrigeration equipment actually shuts down When the duration is greater than the seventh threshold, the second neural network model is trained according to the currently acquired sample data, the sample data including outdoor temperature, indoor temperature and refrigeration equipment load. That is to say, in the case of good environmental conditions (for example, there is a fast Ethernet interconnection between FSU and cloud UME, and cloud UME computing resources are sufficient), real-time or online model training can be supported.
  • the refrigeration equipment control method of the present disclosure may further include: combining the acquired sample data of the day and the actual operation parameters of the refrigeration equipment on the day
  • the training sample data set is added to train the first neural network model and the third neural network model according to the training sample data set.
  • FSU In the case of communication network interruption, FSU cannot communicate with UME. In order to realize the control of refrigeration equipment, FSU can automatically run the built-in temperature start-stop control algorithm, and can also receive and save the cooling control plan issued by UME in advance, and run it locally Refrigeration linkage control algorithm copied from UME.
  • the refrigeration equipment control method of the present disclosure may further include: combining the first neural network model and the second neural network model And the third neural network model is deployed on the FSU, so that when the FSU and the UME fail to communicate, the optimal control parameters of the refrigeration equipment of the day are determined, and the operation of the refrigeration equipment is controlled according to the optimal control parameters.
  • An application scenario of the present disclosure is: a base station type computer room where the heat generated by the communication equipment in the computer room is less than 10 KW, which is usually a data, transmission, and switching type base station computer room of an operator.
  • the original computer room refrigeration equipment only had one air conditioner.
  • an indirect heat exchanger such as intelligent heat pipe equipment (HPE) was installed to pass the air conditioner.
  • HPE intelligent heat pipe equipment
  • the use of heat pipe technology does not require mechanical cooling, and the temperature difference between indoor and outdoor is basically maintained at about 6 degrees, so it can be applied to more than 90% of the year.
  • the energy consumption of its components is much lower than that of the traditional compressor air conditioner, and the energy consumption is only about 1/5 of the original air conditioning system, so it can greatly save the power consumption of the air conditioner.
  • the refrigeration equipment control scheme can be based on big data technology and neural network technology, fully taking into account the current indoor and outdoor temperature and humidity, system load and other data, combined with load prediction, weather forecast, historical sample data of the same period, etc., through neural network Calculating, predicting the load and indoor temperature of the refrigeration equipment in advance, and outputting the optimal plan for the linkage control of the refrigeration equipment on the day, combined with the traditional control rule strategy, can realize the predictable active control of the air conditioning and heat exchange equipment in the computer room to achieve optimization The purpose of control, energy saving and consumption reduction.
  • the air conditioner running time and the number of starts are significantly reduced; at the same time, the working temperature of the equipment in the computer room can be increased to a controllable safety range of 30-40°C, further reducing The energy consumption of refrigeration equipment is improved.
  • the active predictive joint control scheme of air-conditioning and heat exchange equipment can save nearly 10,000 kWh of power consumption for communication base stations each year, and the average power consumption is reduced by 40%. Calculated with a 10% ratio of base stations, annual electricity bills will be reduced by 5 billion yuan and 1.35 million tons of carbon emissions will be reduced, with significant economic and social benefits.
  • the present disclosure also provides a refrigeration equipment control device.
  • FIG 10 and 11 are schematic diagrams of the structure of the refrigeration equipment control device provided by the present disclosure.
  • the refrigeration equipment control device includes a first processing module 101, a second processing module 102, and a control module 103.
  • the first processing module 101 is used to determine the current outdoor temperature.
  • the second processing module 102 is used to: input the historical sample data of the refrigeration equipment load and the preset influence factors as the first input parameters into the first neural network model to obtain the load predicted on the day of the refrigeration equipment; The data and the load predicted by the refrigeration equipment on the day are input as the second input parameters to the second neural network to obtain the predicted indoor temperature on the day; and the predicted indoor temperature on the day and the preset cooling efficiency factor are input as the third input parameters
  • the third neural network obtains the optimal control parameters of the refrigeration equipment of the day.
  • the control module 103 is configured to control the operation of the refrigeration equipment according to the optimal control parameter.
  • the refrigeration equipment control device of the present disclosure may further include a model establishment module 104.
  • the model establishment module 104 is configured to establish the first neural network model, the second neural network model, and the third neural network model in the initialization phase.
  • the model building module 104 may be used to: obtain historical sample data, the sample data including outdoor temperature, indoor temperature, and refrigeration equipment load; simulate and simulate the historical sample data to calculate the daily optimal control parameters of the refrigeration equipment; and According to the historical sample data and the daily optimal control parameters of the refrigeration equipment, a first neural network model, a second neural network model, and a third neural network model are established.
  • the model building module 104 may also be used to: after simulating the historical sample data to calculate the daily optimal control parameters of the refrigeration equipment, and after calculating the daily optimal control parameters of the refrigeration equipment based on the historical sample data and the refrigeration equipment Control parameters, before establishing the first neural network model, the second neural network model, and the third neural network model, normalize the historical sample data and the daily optimal control parameters of the refrigeration equipment; and
  • the transformed data establishes a training sample data set, and the training sample data set includes a training set, a verification set, and a test set.
  • the model establishment module 104 may be used to establish a first neural network model, a second neural network model, and a third neural network model according to the training sample data set.
  • the model building module 104 may be used to: use the historical sample data of the refrigeration equipment load and the preset influence factor as the first input parameter, and use the historical sample data of the refrigeration equipment load of the day as the first output parameter to establish The first neural network model; the historical sample data of the same period of outdoor temperature and the historical sample data of the day load of the refrigeration equipment are used as the second input parameter, and the historical sample data of the indoor temperature of the day is used as the second output parameter to establish the second nerve Network model; and the historical sample data of the indoor temperature of the day and the preset cooling efficiency factor are used as the third input parameter, and the historical sample data of the optimal control parameter of the refrigeration equipment of the day is used as the third output parameter to establish the first Three neural network model.
  • the sample data may include analog data and sampling data.
  • the simulated data is data obtained by simulating the operation of the refrigeration equipment when the indoor temperature is greater than the preset third threshold.
  • the sampled data is data sampled when the indoor temperature is less than the preset sixth threshold and the actual shutdown duration of the refrigeration equipment is greater than the preset seventh threshold.
  • the first processing module 101 may be used to determine the outdoor temperature within a preset period of time before the current moment; and according to the outdoor temperature within the preset period of time before the current moment, the predicted temperature of the day, and preset first and second weights Determine the current outdoor temperature.
  • the optimal control parameters may include the turn-on time and the turn-on duration.
  • the impact factor may include one or any combination of the following: holiday impact factor, tide impact factor, and regional event factor.
  • the control module 103 may be used to: if the current indoor temperature is less than or equal to a preset first threshold and greater than or equal to a preset second threshold, and meets the first high-temperature pre-start condition, set the maximum operating time of the air conditioner to the air conditioner on And start the air conditioner, the second threshold is less than the first threshold; and if the actual air conditioner on time is greater than or equal to the maximum operating time of the air conditioner, the air conditioner is turned off.
  • the first high-temperature pre-start condition may include: reaching the time when the air conditioner is turned on, the current indoor temperature is greater than a preset third threshold, and the actual shutdown duration of the air conditioner is greater than the preset minimum shutdown duration of the air conditioner.
  • the control module 103 may also be used to: in the process of controlling the operation of the refrigeration equipment according to the optimal control parameters, if the current indoor temperature is greater than the first threshold and the actual shutdown duration of the air conditioner is greater than the shortest shutdown duration of the air conditioner, then The maximum operating time of the air conditioner is set to the maximum on time of the air conditioner, and the air conditioner is started; and/or, if the current indoor temperature is less than the second threshold, the air conditioner is turned off.
  • the control module 103 may also be used to: if the second high-temperature pre-start condition is met, start the heat exchange device; and if the actual opening time of the heat exchange device is greater than or equal to the opening time of the heat exchange device, turn off the heat exchange equipment.
  • the second high-temperature pre-start condition may include: reaching the time when the heat exchange equipment is turned on, and the current indoor temperature is greater than a preset fourth threshold, and the current indoor temperature and outdoor temperature The difference between is greater than the preset fifth threshold.
  • the second high-temperature pre-start condition may include one of the following: the time when the heat exchange device is turned on, and the current indoor temperature is greater than the preset fourth threshold, and the current indoor temperature The difference with the outdoor temperature is greater than the preset eighth threshold, the eighth threshold is greater than the fifth threshold; and the time when the heat exchange device is turned on, and the current indoor temperature is greater than the preset fourth threshold, and the current The difference between the indoor temperature and the outdoor temperature is greater than the preset eighth threshold, and the current indoor humidity is less than or equal to the preset ninth threshold.
  • the control module 103 may also be used to: if the air conditioner is turned on, the heat exchange equipment is turned off; If the heat exchange device is turned on, the air conditioner is turned off.
  • the control module 103 may also be used to: after controlling the operation of the refrigeration equipment according to the optimal control parameters, if the error between the actual operating parameters of the air conditioner on that day and the optimal control parameters of the air conditioner on that day exceeds the preset tenth threshold , The second processing module 102 is instructed to re-determine the optimal control parameters of the day of the air conditioner; and update the training sample data set according to the re-determined optimal control parameters of the day of the air conditioner.
  • the control module 103 can also be used to: if one type of refrigeration equipment currently running is faulty and the other type of refrigeration equipment is normal, turn off the failed refrigeration equipment and turn on the normal refrigeration equipment; and if the two currently running refrigeration equipment is normal If the refrigeration equipment is faulty, when the fault is eliminated, the refrigeration equipment with the fault eliminated will be activated.
  • the second processing module 102 may also be used to train the second nerve according to the currently acquired sample data if the current indoor temperature is less than the preset sixth threshold and the actual shutdown duration of the refrigeration equipment is greater than the preset seventh threshold.
  • the network model, the sample data includes outdoor temperature, indoor temperature and refrigeration equipment load.
  • the present disclosure also provides a computer device that includes one or more processors and a storage device.
  • the storage device stores one or more programs. When the one or more programs are processed by the one or more When the device is executed, the one or more processors implement the refrigeration device control method provided in the present disclosure.
  • the present disclosure also provides a computer-readable medium on which a computer program is stored.
  • the processor realizes the refrigeration device control method provided in the present disclosure.
  • Such software may be distributed on a computer-readable medium
  • the computer-readable medium may include a computer storage medium (or non-transitory medium) and a communication medium (or transitory medium).
  • the term computer storage medium includes volatile and non-volatile data implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data).
  • Information such as computer-readable instructions, data structures, program modules, or other data.
  • Computer storage media include but are not limited to RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, digital versatile disk (DVD) or other optical disk storage, magnetic cassettes, magnetic tapes, magnetic disk storage or other magnetic storage devices, or Any other medium used to store desired information and that can be accessed by a computer.
  • a communication medium usually contains computer-readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transmission mechanism, and may include any information delivery medium. .

Landscapes

  • Engineering & Computer Science (AREA)
  • Combustion & Propulsion (AREA)
  • General Engineering & Computer Science (AREA)
  • Mechanical Engineering (AREA)
  • Chemical & Material Sciences (AREA)
  • Signal Processing (AREA)
  • Fuzzy Systems (AREA)
  • Mathematical Physics (AREA)
  • Physics & Mathematics (AREA)
  • Human Computer Interaction (AREA)
  • Health & Medical Sciences (AREA)
  • Biomedical Technology (AREA)
  • Air Conditioning Control Device (AREA)
  • Feedback Control In General (AREA)

Abstract

本公开提供一种制冷设备控制方法，包括：确定当前室外温度；将制冷设备负荷的同期历史样本数据和预设的影响因子作为第一输入参数输入第一神经网络模型，得到制冷设备当日预测的负荷；将室外温度的同期历史样本数据和制冷设备当日预测的负荷作为第二输入参数输入第二神经网络，得到当日预测的室内温度；将当日预测的室内温度和预设的制冷效率因子作为第三输入参数输入第三神经网络，得到制冷设备当日的最优控制参数；以及根据最优控制参数控制制冷设备运行。本公开还提供一种制冷设备控制装置、计算机设备和计算机可读介质。

Description

制冷设备控制方法、装置、计算机设备和计算机可读介质 技术领域
本公开涉及自动控制技术领域，具体涉及一种制冷设备控制方法、装置、计算机设备和计算机可读介质。
背景技术
在移动通信网络中，约80％的能耗来自广泛分布的基站，越加密集的基站意味着更高的能耗。通常情况下，基站机房会根据机房类型(砖瓦房、房舱、彩钢板房等)和机房内设备负荷来选择不同容量的空调，对过热的设备和装置降温，保障设备安全运行。移动基站耗能占比中，空调制冷是最大一块，优化空调控制算法，是移动通信基站降低能耗、减少电费开支最重要的努力方向之一。
发明内容
本公开提供一种制冷设备控制方法，包括：确定当前室外温度；将制冷设备负荷的同期历史样本数据和预设的影响因子作为第一输入参数输入第一神经网络模型，得到制冷设备当日预测的负荷；将室外温度的同期历史样本数据和所述制冷设备当日预测的负荷作为第二输入参数输入第二神经网络，得到当日预测的室内温度；将所述当日预测的室内温度和预设的制冷效率因子作为第三输入参数输入第三神经网络，得到所述制冷设备当日的最优控制参数；以及根据所述最优控制参数控制所述制冷设备运行。
本公开还提供一种制冷设备控制装置，包括：第一处理模块、第二处理模块和控制模块，所述第一处理模块用于确定当前室外温度；所述第二处理模块用于：将制冷设备负荷的同期历史样本数据和预设的影响因子作为第一输入参数输入第一神经网络模型，得到制冷设备当日预测的负荷；将室外温度的同期历史样本数据和所述制冷设备当日预测的负荷作为第二输入参数输入第二神经网络，得到当日预测的室内温度；将所述当日预测的室内温度和预设的制冷效率因子作为第 三输入参数输入第三神经网络，得到所述制冷设备当日的最优控制参数；所述控制模块用于根据所述最优控制参数控制所述制冷设备运行。
本公开还提供一种计算机设备，包括：一个或多个处理器；以及存储装置，其上存储有一个或多个程序；当所述一个或多个程序被所述一个或多个处理器执行时，使得所述一个或多个处理器实现根据本公开的制冷设备控制方法。
本公开还提供一种计算机可读介质，其上存储有计算机程序，其中，所述程序被处理器执行时，使得所述处理器实现根据本公开的制冷设备控制方法。
附图说明
图1为本公开提供的制冷设备控制系统的示意图；
图2至图4为本公开提供的建立第一、二、三神经网络模型的流程示意图；
图5为本公开提供的制冷设备控制流程示意图；
图6A至图6C为本公开提供的第一、二、三神经网络模型示意图；
图7为本公开提供的空调控制流程示意图；
图8为本公开提供的换热设备控制流程示意图；
图9为本公开提供的再次确定并更新制冷设备当日的最优控制参数的流程示意图；以及
图10和图11为本公开提供的制冷设备控制装置的结构示意图。
具体实施方式
在下文中将参考附图更充分地描述示例实施例，但是所述示例实施例可以以不同形式来体现且不应当被解释为限于本文阐述的实施例。提供这些实施例的目的在于使本公开透彻和完整，并将使本领域技术人员充分理解本公开的范围。
如本文所使用的，术语“和/或”包括一个或多个相关列举条目的任何和所有组合。
本文所使用的术语仅用于描述特定实施例，且不意欲限制本公开。如本文所使用的，单数形式“一个”和“该”也意欲包括复数形式，除非上下文另外清楚指出。还将理解的是，当本说明书中使用术语“包括”和/或“由……制成”时，指定存在所述特征、整体、步骤、操作、元件和/或组件，但不排除存在或添加一个或多个其他特征、整体、步骤、操作、元件、组件和/或其群组。
本文所述实施例可借助本公开的理想示意图而参考平面图和/或截面图进行描述。因此，可根据制造技术和/或容限来修改示例图示。因此，实施例不限于附图中所示的实施例，而是包括基于制造工艺而形成的配置的修改。因此，附图中例示的区具有示意性属性，并且图中所示区的形状例示了元件的区的具体形状，但并不旨在是限制性的。
除非另外限定，否则本文所用的所有术语(包括技术和科学术语)的含义与本领域普通技术人员通常理解的含义相同。还将理解，诸如那些在常用字典中限定的那些术语应当被解释为具有与其在相关技术以及本公开的背景下的含义一致的含义，且将不解释为具有理想化或过度形式上的含义，除非本文明确如此限定。
运营商在新建和扩建基站机房时，会考虑配置节能型换热设备，替代空调或者与空调进行联动控制，以保障机房内各类设备长期稳定的工作条件。节能型换热设备的基本原理是以室外的自然环境为冷源，当室外空气温度低于室内温度一定程度时，将室外低温空气与机房内高温空气进行热交换，把机房的热量带走，达到降低机房温度的目的，从而减少空调设备的使用时间，节约电能。
常规联动控制算法是传统的温控启停方法，以环境温度作为换热设备和空调联动控制的主要依据，算法简单但难以改进。常规联动控制过程如下：实时检测室内外温度，如果室内温度超过设备运行的温度上限，则启动换热设备或空调制冷：当换热设备启动条件满足时(如室内外温差达到阈值)，优先启动换热设备，否则启动空调。空调和换热设备不宜频繁切换，间隔半小时以上。
分别设置换热设备和空调的启停条件参数，比如换热设备的启 停温度可为35/25℃，温差为8℃，即，室温超过35℃时开启换热设备，室温低于25℃时停机，室内外温差超过8℃时允许启动换热设备。但实际工程应用中，启停条件参数通常很难确定，而且，针对不同地域、不同季节性气候、不同的早晚温差，启停条件参数不是固定的，如果设置固定的启停条件参数，会造成空调的频繁启动，增加能耗。常规联动控制算法仅考虑了环境温度这个外部因素，空调的启动时间和启动次数是不可预期的，且控制精度低，想要改进非常困难。
四季气候轮转、气温变化、负载变化、设备实际温度等各种因素，以及这些因素的变化组合，它们都对制冷设备运行策略产生影响，因此，制冷设备控制策略缺乏可循的规律。以换热设备启动温度为35℃、空调启动温度为40℃为例，假设某基站室温超过35℃的时候不多，平常开启换热设备足以满足热负荷的要求，无需开启空调，但在某个业务高峰和气温高峰的叠加期，室温偶尔会超过40℃。按照传统控制算法，需要开启空调，但如果能提前预知40℃以上高温时间将会很短暂、不会影响设备安全运行的话(部分基站/传输设备工作范围长期可到40℃，短时50℃)，实际上不需要开启空调。这样，在保障设备安全的同时，可以避免一次空调的开启，实现了一定程度的节能。
本公开提供一种制冷设备控制方法，所述方法可以控制机房内制冷设备的运行。所述方法可应用于图1所示的制冷控制系统，如图1所示，本公开提供的制冷控制系统包括制冷设备控制装置、现场控制器(Field Supervision Unit，FSU)和制冷设备。FSU为现场设备，设置于制冷设备所在的机房，包括采集单元和执行单元。采集单元用于采集室外温度和湿度、室内温度、设备负荷等实时数据，并上传给制冷控制装置。执行单元用于根据制冷控制装置的指示控制制冷设备运行。制冷设备控制装置可为云端设备，并且可以选用统一管理专家(Unified Management Expert，UME)，其上配置有第一神经网络(Neural Network，NN)模型、第二神经网络模型、第三神经网络模型、历史样本数据库以及制冷设备的控制策略(例如，制冷控制算法)。UME可以根据FSU上报的数据和第一神经网络模型、第二神经 网络模型、第三神经网络模型得到制冷设备的预测控制方案，并将预测控制方案下发给FSU。制冷设备可以包括空调和换热设备，可以根据下发的控制方案运行。
在初始化阶段，可以在制冷设备控制装置内预设以下第一阈值至第十阈值以及各种时长。
第一阈值VHT，例如可以为45℃，当室内温度超过VHT时，空调无条件启动。第二阈值VLT，例如可以为15℃，当室内温度低于VLT时，空调无条件关闭，第二阈值VLT小于第一阈值VHT。第三阈值HT AC，例如可以为40℃，当室内温度超过HT AC时，可以启动空调。第四阈值HT HEE，例如可以为35℃，当室内温度超过HT HEE时，可以启动换热设备。第五阈值，用于判断是否满足间接换热设备的第二高温预启动条件。第六阈值LT，例如可以为25℃，当室内温度低于LT时，可以关闭空调和换热设备，第六阈值LT小于第四阈值HT HEE和第三阈值HT AC。第七阈值，用于判断制冷设备停机时长。第八阈值，用于判断是否满足直接换热设备的第二高温预启动条件中的室内外温差。第九阈值，用于判断是否满足直接换热设备的第二高温预启动条件中的湿度。第十阈值，用于判断空调实际运行参数与空调最优控制参数之间的误差。空调最大开启时长MAXCOT和空调最短停机时长MINCST，通常，空调最大开启时长MAXCOT为12小时，空调最短停机时长MINCST为0.5小时。
在初始化阶段建立第一神经网络模型、第二神经网络模型和第三神经网络模型。以下结合图2，对建立第一神经网络模型、第二神经网络模型和第三神经网络模型的流程进行详细说明。
如图2所示，所述建立第一神经网络模型、第二神经网络模型和第三神经网络模型包括以下步骤S21至S23。
在步骤S21，获取历史样本数据。
样本数据可以包括室外温度、室内温度和制冷设备负荷。
在步骤S21中，制冷设备控制装置可以从历史数据库中获取历史样本数据。历史数据库中可以存储大量的每日室外温度T Rout、室内温度T Rin、制冷设备负荷L R等历史样本数据。可以根据这些参数变化 缓急程度确定采样周期，例如，室外温度T Rout的采样周期可以为10分钟，室内温度T Rin和制冷设备负荷L R的采样周期可以为5分钟。
样本数据可以包括模拟数据和采样数据。模拟数据是当室内温度大于第三阈值HT AC时，经过模拟制冷设备的运行得到的数据。采样数据是当室内温度小于第六阈值LT且制冷设备实际停机时长大于第七阈值时采样得到的数据。也就是说，在室内温度T Rin较高、需要制冷设备运行的情况下，可以利用假负载模拟真实制冷设备，并记录T Rout、T Rin、L R等数据。在室内温度T Rin较低并且制冷设备较长停机的情况下(比如室外温度T Rout较低的季节或夜间)，可以直接使用大量已有的历史样本数据，以加快历史样本数据采集速度。
在步骤S22，对历史样本数据仿真模拟，计算得到制冷设备每日的最优控制参数。
在步骤S22中，通过计算机仿真训练，建立机房环境、发热设备和制冷设备的热分布图，对历史样本数据模拟计算，输出制冷设备当日控制最优解向量(即，制冷设备每日的最优控制参数)，并将定制冷设备每日的最优控制参数保存为样本数据标签。
根据仿真结果和日常经验，空调不宜频繁开启，例如，可以限定每日空调最多开启12次，每日换热设备最多开启12次。也就是说，对于空调来说，若一个开启时刻/开启时长(T moment/T hours)标签组有2个有效值，则意味着当日空调最优控制参数是：当日开启空调运行两次，每次在开启时刻T moment开启，运行时长为对应的开启时长T hours的值。
在步骤S23，根据历史样本数据和制冷设备每日的最优控制参数，建立第一神经网络模型、第二神经网络模型和第三神经网络模型。
在步骤S23中，依次建立第一神经网络模型、第二神经网络模型和第三神经网络模型。
如图3所示，在对历史样本数据仿真模拟，计算得到制冷设备每日的最优控制参数的步骤(即，步骤S22)之后，并且在根据历史样本数据和制冷设备每日的最优控制参数，建立第一神经网络模型、第二神经网络模型和第三神经网络模型的步骤(即，步骤S23)之前， 本公开的制冷设备控制方法还可以包括步骤S22'至S23'。
在步骤S22'，对历史样本数据和制冷设备每日的最优控制参数进行归一化处理。
可以根据以下公式，对历史样本数据和制冷设备每日的最优控制参数进行归一化处理，使数据处于范围(0，1)之间：
Figure PCTCN2021099313-appb-000001
其中，X real为实际样本的真实值，X *为归一化处理后的数据，X max为对应类型数据样本的最大值或上限值，X min为对应类型数据样本的最小值或下限值。
对于室外温度T Rout和室内温度T Rin，X max可以为上限值100℃，X min可以为下限值-40℃，因此各真实温度X real的归一化值X *＝(X real-X min)/(X max-X min)＝(X real+40)/140。对于制冷设备负荷L R，设定X max可以为制冷设备的满负荷，X min可以为0，因此制冷设备负荷L R的X real归一化值X *＝X real/X max。
对于空调开启时刻T moment-AC和换热设备开启时刻T moment-HEE(其格式可以为hh:mm:ss)，可以设定X max为上限值1440(即，一天有24*60＝1440分钟)，X min可以为0，因此T moment-AC和T moment-HEE的归一化值X *＝(hh*60+mm)/1440。对于空调开启时长T hours-AC和换热设备开启时长T hours-HEE，可以设定X max为上限值24(即，一天有24小时)，X min可以为0，因此T hours-AC和T hours-HEE的归一化值X *＝X real/24。
在步骤S23'，根据归一化处理后的数据建立训练样本数据集，训练样本数据集包括训练集、验证集和测试集。
在步骤S23'中，可以按照6:2:2的样本比例建立训练集、验证集和测试集。
相应地，根据历史样本数据和制冷设备每日的最优控制参数，建立第一神经网络模型、第二神经网络模型和第三神经网络模型的步骤(即，步骤S23)可以包括：根据训练样本数据集建立第一神经网络模型、第二神经网络模型和第三神经网络模型。
以下结合图4，对建立第一神经网络模型、第二神经网络模型和第三神经网络模型的流程进行详细说明。
如图4所示，根据历史样本数据和制冷设备每日的最优控制参数，建立第一神经网络模型、第二神经网络模型和第三神经网络模型的步骤(即，步骤23)可以包括步骤S231至S233。
在步骤S231，将制冷设备负荷的同期历史样本数据和预设的影响因子作为第一输入参数，并将制冷设备当日负荷的历史样本数据作为第一输出参数，建立第一神经网络模型。
影响因子可以包括以下之一或任意组合：节假日影响因子F holiday、潮汐影响因子F tide、区域事件因子F event。节假日影响因子F holiday、潮汐影响因子F tide和区域事件因子F event的取值范围均为(0，1)，并且可以根据人工经验确定。例如，针对居民小区而言，正常工作日的节假日影响因子F holiday可以为0、双休日的节假日影响因子F holiday可以为0.1、春节长假的节假日影响因子F holiday可以为0.25等；针对工业园而言，工作时间段的潮汐影响因子F tide可以为0.5、加班时间段的潮汐影响因子F tide可以为0.7、深夜时间段的潮汐影响因子F tide可以为0.3等；针对某些区域而言，正常区域事件因子F event可以为0，有商业营销活动的区域事件因子F event可以为0.1，集会的区域事件因子F event可以为0.2，演唱会的区域事件因子F event可以为0.3等。
在步骤S232，将室外温度的同期历史样本数据和制冷设备当日负荷的历史样本数据作为第二输入参数，并将当日室内温度的历史样本数据作为第二输出参数，建立第二神经网络模型。
在步骤S233，将当日室内温度的历史样本数据和预设的制冷效率因子作为第三输入参数，并将制冷设备当日最优控制参数的历史样本数据作为第三输出参数，建立第三神经网络模型。
最优控制参数可以包括开启时刻T moment和开启时长T hours，即，空调开启时刻T moment-AC、换热设备开启时刻T moment-TEE，空调开启时长T hours-AC和换热设备开启时长T hours-TEE。
制冷效率因子可以包括换热制冷效率因子F eff1和空调制冷效率因子F eff2。当机房环境固定时，换热制冷效率因子F eff1和空调制冷效率因子F eff2都是常量，若机房环境发生改变(比如，制冷设备更换或空间位置挪动等)，则需将换热制冷效率因子F eff1和空调制冷效率因 子F eff2调整为新的常量。
假设换热设备的24组T moment/T hours数据中有2个有效值，比如T moment1为0.45、T hours1为0.05和T moment2为0.60、T hours2为0.10，而空调的12组T moment/T hours都没有有效值。将开启时刻T moment回转到hh:mm:ss格式，并且将开启时长T hours回转到标准时长后，制冷设备当日最优控制参数的含义如下：
(1)当日换热设备预开启运行两次；
(2)换热设备第一次开启时刻为10:48(即，0.45*24＝10.8＝10:48)，并运行1.2小时(即，0.05*24＝1.2)，即，运行时间区间为10:48至12:00(即，0.45*24+0.05*24＝12＝12:00)；
(3)换热设备第二次开启时刻为14:24(即，0.60*24＝14.4＝14:24)，并运行2.4小时(即，0.10*24＝2.4)，即，运行时间区间为14:24至16:48(即，0.60*24+0.10*24＝16.8＝16:48)；
(4)当日空调不开启运行。
第一神经网络模型、第二神经网络模型和第三神经网络模型经过训练和优化后，可以根据实际运行环境进行部署。三个神经网络模型可以都部署在UME上，以充分利用云端的强大算力资源，从而实现实时或在线的训练。如有必要，也可以通过增加计算棒等方式，将三个神经网络模型部署在边缘侧，例如，部署在FSU上。
图5为本公开提供的制冷设备控制流程示意图。
如图5所示，本公开提供的制冷设备控制方法可以用于控制制冷设备运行，并且包括步骤S11至S15。
在步骤S11，确定当前室外温度。
可以根据预测温度和检测的室外温度经过加权计算得到当前室外温度T Rout，即，先确定当前时刻前预设时长内的室外温度，然后根据当前时刻前预设时长内的室外温度、当天预测温度和预设的第一权重和第二权重确定当前室外温度T Rout。通常，预设时长可以为1小时，当天预测温度可以是天气预报预测的当天温度。例如，室外温度T Rout＝天气预报温度*0.8+上一小时实测室外温度*0.2。
需要说明的是，FSU可以采集室内外温度、湿度、制冷设备负荷 等数据并上传给UME。
在步骤S12，将制冷设备负荷的同期历史样本数据和预设的影响因子作为第一输入参数输入第一神经网络模型，得到制冷设备当日预测的负荷。
同期是指历史上同一时期，例如，去年今天的同一时刻、前年今天的同一时刻均可以为今天该时刻的同期。
在步骤S12中，如图6A所示，将制冷设备负荷的同期历史样本数据L N和节假日影响因子F holiday、潮汐影响因子F tide和区域事件因子F event输入第一神经网络模型，得到制冷设备当日预测的负荷L R，作为第一神经网络模型的输出值。
在步骤S13，将室外温度的同期历史样本数据和制冷设备当日预测的负荷作为第二输入参数输入第二神经网络，得到当日预测的室内温度。
在步骤S13中，如图6B所示，将将室外温度的同期历史样本数据T Rout和制冷设备当日预测的负荷L R(即，第一神经网络的输出值)输入第二神经网络模型，得到当日预测的室内温度T Rin，作为第二神经网络模型的输出值。
在步骤S14，将当日预测的室内温度和预设的制冷效率因子作为第三输入参数输入第三神经网络，得到制冷设备当日的最优控制参数。
在步骤S14中，如图6C所示，将当日预测的室内温度T Rin(即，第二神经网络的输出值)、换热制冷效率因子F eff1和空调制冷效率因子F eff2输入第三神经网络模型，得到空调当日的最优控制参数(即，空调开启时刻T moment-AC和空调开启时长T hours-AC)和换热设备当日的最优控制参数(即，换热设备开启时刻T moment-TEE和换热设备开启时长T hours-TEE)。
需要说明的是，在实际使用时，可以将开启时刻T moment回转到hh:mm:ss格式，并且将开启时长T hours回转到标准时长(例如，xx小时)。
在步骤S12至S14中，UME依次运行第一神经网络模型、第二神经网络模型和第三神经网络模型，以输出制冷设备当日的最优控制参 数。
空调的最优控制参数可以包括空调开启时刻T moment-AC和空调开启时长T hours-AC，换热设备的最优控制参数可以包括换热设备开启时刻T moment-TEE和换热设备开启时长T hours-TEE。
最优控制参数可以包括每日至多12组空调开启时刻T moment-AC和空调开启时长T hours-AC，以及至多24组换热设备开启时刻T moment-TEE和换热设备开启时长T hours-TEE。
在步骤S15，根据最优控制参数控制制冷设备运行。
在步骤S15中，根据空调的最优控制参数控制空调运行，并且根据换热设备的最优控制参数控制换热设备运行。
根据本公开提供的制冷设备控制方法，利用神经网络模型结合当前室外温度、制冷设备负荷的同期历史样本数据、影响因子和制冷效率因子等参数，实现对空调和换热设备的控制方案预测及联动控制，预测得到的控制方案具有较高的精度，克服了传统算法难改进的缺陷，实现了对空调和换热设备主动控制，优化了运行效率，降低能耗；另外，将历时数据与当前实测数据相结合，并考虑到特殊事件的影响因素以及制冷设备的制冷效率的影响因素，使得预测得到的控制方案更为准确，能够适应机房环境变化，提升应用范围。
图7为本公开提供的空调控制流程示意图。
如图7所示，本公开提供的空调控制流程包括步骤S31至S39。
在步骤S31，若当前室内温度大于第一阈值VHT，则执行步骤S36；否则，执行步骤S32。
在步骤S31中，若当前室内温度大于VHT，说明当前室内温度过高，则可以进一步判断空调运行是否超时(即，执行步骤S36)；若当前室内温度小于或等于VHT，则可以进一步判断当前室内温度是否过低(即，执行步骤S32)。
在步骤S32，若当前室内温度小于第二阈值VLT，则执行步骤S39；否则，执行步骤S33。
在步骤S32中，若当前室内温度小于第二阈值VLT，说明当前室内温度过低，则空调可以因低温异常而停机(即，执行步骤S39)； 若当前室内温度大于或等于第二阈值VLT，说明当前室内温度不会因高温异常而停机也不会因低温异常而停机，则可以进一步判断是否满足第一高温预启动条件(即，执行步骤S33)。
在步骤S33，若满足第一高温预启动条件，则执行步骤S34；否则，返回步骤S31。
在步骤S33中，当前室内温度小于或等于第一阈值VHT且大于或等于第二阈值VLT，若满足第一高温预启动条件，则根据空调当日的最优控制参数控制空调运行(即，执行步骤S34)；若不满足第一高温预启动条件，则返回步骤S31。
第一高温预启动条件可以包括：到达空调开启时刻T moment-AC，且当前室内温度大于第三阈值HT AC，且空调实际停机时长大于空调最短停机时长MINCST。
在步骤S34，将空调最大运行时长T on-max设置为空调开启时长T hours-AC和空调最大开启时长MAXCOT中的最小值。
在步骤S34中，可以取空调开启时长T hours-AC和空调最大开启时长MAXCOT中的最小值作为实际控制空调运行的控制参数，以保证空调运行的可靠性和安全性。
步骤S35，启动空调，并执行步骤S38。
在步骤S35中，控制空调启动后，开始记录空调实际开启时长T on-AC，将空调实际停机时长T off-AC清零，并执行步骤S38。
在步骤S36，若空调实际停机时长T off-AC大于空调最短停机时长MINCST，则执行步骤S37；否则，返回步骤S31。
在步骤S36中，当前室内温度大于第一阈值VHT，若当前空调实际停机时长T off-AC大于空调最短停机时长MINCST，说明满足高温异常启动条件，则执行空调高温异常启动操作(即，执行步骤37)；若当前空调实际停机时长T off-AC小于或等于空调最短停机时长MINCST，则返回步骤S31。
在步骤S37，将空调最大运行时长T on-max设置为空调最大开启时长MAXCOT，并执行步骤S35。
在步骤S37中，在空调高温异常启动的情况下，直接根据预设 的空调最大开启时长MAXCOT控制空调的运行时长。
在步骤S38，若空调实际开启时长T on-AC大于或等于空调最大运行时长T on-max，则执行步骤S39；否则，保持空调当前状态。
空调启动之后，开始记录空调实际开启时长T on-AC，若空调实际开启时长T on-AC大于或等于空调最大运行时长T on-max，则关闭空调；否则，保持空调当前状态。
在步骤S39，关闭空调，并返回步骤S31。
在步骤S39中，控制空调关闭后，开始记录空调实际停机时长T off-AC，并将空调实际开启时长T on-AC清零，随后返回步骤S31继续检测室内温度。
空调控制流程还可以包括：若当前室内温度小于第六阈值LT，则关闭空调。
通过上述步骤S31至S39可以看出，在应用第三神经网络模型输出的预测方案的基础上，结合预置的空调和换热设备启停策略的算法，可保障在神经网络模型预测异常时，空调和换热设备也能安全运行。在实际室温超过第一阈值VHT时，空调可以因高温异常而启动；在实际室温低于第二阈值VLT时，空调可以因低温异常而停机；在达到空调开启时刻T moment-AC且实际室温超过第三阈值HT AC且满足运行间隔超过最短停机时长MINCST时，空调将按照第三神经网络模型输出的预测方案运行，即，在空调开启时刻T moment-AC到达时启动运行，运行持续时间为空调开启时长T hours-AC。
图8为本公开提供的换热设备控制流程示意图.
如图8所示，本公开提供的换热设备控制流程包括步骤S41至S44。
在步骤S41，若满足第二高温预启动条件，则执行步骤S42；否则，保持换热设备当前的状态。
需要说明的是，换热设备可以包括直接换热设备和间接换热设备，直接换热设备可以包括新风系统，间接换热设备可以包括热管设备(Heat Pipe Equipment，HPE)。
当换热设备为间接换热设备时，第二高温预启动条件可以包括： 到达换热设备开启时刻T moment-TEE，且当前室内温度大于第四阈值HT HEE，且当前室内温度与室外温度的差值大于第五阈值。
当换热设备为直接换热设备时，第二高温预启动条件包括以下之一：
(1)到达换热设备开启时刻T moment-TEE，且当前室内温度大于第四阈值HT HEE，且当前室内温度与室外温度的差值大于第八阈值。第八阈值可以大于第五阈值，也就是说，在第二高温预启动条件中，直接换热设备的室内外温差要求高于间接换热设备的室内外温差要求，例如，第五阈值可以为6℃，第八阈值可与为10℃。
(2)到达换热设备开启时刻T moment-TEE，且当前室内温度大于第四阈值HT HEE，且当前室内温度与室外温度的差值大于第八阈值，且当前室内湿度小于或等于第九阈值。也就是说，直接换热设备的第二高温预启动条件可以包括温度条件和湿度条件，例如，第九阈值可以为90％。
在步骤S42，启动换热设备。
在步骤S42中，控制换热设备启动后，开始记录换热设备实际开启时长T on-HEE，并将换热设备实际停机时长T off-HEE清零。
在步骤S43，若换热设备实际开启时长T on-HEE大于或等于换热设备开启时长T hours-HEE，则执行步骤S44；否则，保持换热设备当前的状态。
在步骤S44，关闭换热设备。
在步骤S44中，控制换热设备关闭后，开始记录换热设备实际停机时长T off-HEE，并将换热设备实际开启时长T on-HEE清零。
换热设备控制流程还可以包括：若当前室内温度小于第六阈值LT，则关闭换热设备。
需要说明的是，空调和间接换热设备可以同时运行，但空调和直接换热设备的运行是互斥的，即，空调和直接换热设备二者不能同时运行。另外，如果出现延误、火情告警时，需要立刻停止运行直接换热设备、关闭风阀，以保障安全。
当换热设备为直接换热设备时，空调开启时刻与换热设备开启 时刻不同；相应地，本公开的制冷设备控制方法还可以包括：若开启空调，则关闭换热设备；若开启换热设备，则关闭空调。
需要说明的是，空调和换热设备控制算法可以运行在UME云端，如有需要，也可以将空调和换热设备控制算法复制到FSU上，实现本地执行，在此情况下，UME要把第三神经网络预测的制冷控制方案提前下发给FSU。
需要说明的是，空调控制和换热设备控制可并发执行，图5所示的步骤S11至S14每日零点前执行一次，输出制冷设备当日的最优控制参数。
图9为本公开提供的再次确定并更新制冷设备当日的最优控制参数的流程示意图。
如图9所示，在根据最优控制参数控制制冷设备运行的步骤(即，图5所示的步骤S15)之后，本公开的制冷设备控制方法还可以包括步骤S51至S53。
在步骤S51，若空调当日的实际运行参数与空调当日的最优控制参数之间的误差超过第十阈值，则执行步骤S52；否则，结束本流程。
在步骤S52，再次确定空调当日的最优控制参数。
步骤S52的具体实现方式与步骤12至S14相同，在此不再赘述。
在步骤S53，根据再次确定的空调当日的最优控制参数更新训练样本数据集。
例如，若当日空调实际开启时刻和当日空调最优控制参数中空调开启时刻之间的误差超过10分钟，则需要重新预测空调当日的最优控制参数，并根据再次预测的空调当日的最优控制参数更新训练样本数据集，以提高制冷控制策略的及时应变能力，以及预测控制的实时性和精度。
神经网络模型可以部署和运行在云端，在外部参数不断变化时，这些模型还可以不断进行实时或在线的训练，以不断提升预测精度，并能适应机房环境改变等异常情况而进行训练和调整。
空调和换热设备可以实现故障倒换，相应地，本公开的制冷设备控制方法还可以包括：若当前运行的一种制冷设备故障且另一种制 冷设备正常，则关闭故障的制冷设备，并开启所述正常的制冷设备；以及若当前运行的两种制冷设备均故障，则在故障消除时，启动故障消除的制冷设备。也就是说，如果当前开启的制冷设备故障，则关闭该故障的制冷设备，并开启正常的制冷设备，等到故障消除时，再开启该制冷设备，并关闭另一个制冷设备。通过空调和换热器在故障时的互为备份的启动运行，可以避免机房异常高温的危险发生。
在根据最优控制参数控制制冷设备运行(即，图5所示的步骤S15)的过程中，本公开的制冷设备控制方法还可以包括：若当前室内温度小于第六阈值LT且制冷设备实际停机时长大于第七阈值，则根据当前获取的样本数据训练第二神经网络模型，所述样本数据包括室外温度、室内温度和制冷设备负荷。也就是说，在环境条件较好(例如，FSU和云端UME间有快速以太网互联，云端UME算力资源充足)的情况下，可以支持实时或在线模型训练。在机房内温度较低、制冷设备较长时间未运行时(比如，气温凉爽的季节、或者低温的夜晚)，可以根据实时采集的室外温度、设备负荷、室内温度等数据，实时在线训练第二神经网络模型。
在根据最优控制参数控制制冷设备运行(即，图5所示的步骤15)之后，本公开的制冷设备控制方法还可以包括：将获取到的当日的样本数据和制冷设备当日的实际运行参数加入训练样本数据集，以便根据训练样本数据集训练第一神经网络模型和第三神经网络模型。通过将当日的样本数据和实际制冷控制结果添加到大数据集中，可以丰富训练集和测试集，对第一和第三神经网络模型进行在线训练，以提升模型的预测精度。
在通信网络中断情况下，FSU无法与UME通信，为了实现制冷设备的控制，FSU可以自动运行内置的温度启停控制算法，也可以接收并保存UME提前下发的制冷控制预案，并在本地运行从UME上复制的制冷联动控制算法。
相应地，在初始化阶段建立了第一神经网络模型、第二神经网络模型和第三神经网络模型之后，本公开的制冷设备控制方法还可以包括：将第一神经网络模型、第二神经网络模型和第三神经网络模型 部署在FSU上，以使在FSU与UME通信故障时，确定制冷设备当日的最优控制参数，并根据所述最优控制参数控制制冷设备运行。
本公开的一个应用场景是：机房通信设备发热量小于10KW的基站类机房，其通常为运营商的数据、传输、交换类基站机房。原先机房制冷设备只有一台空调，为了降低空调能耗，在考虑了基站机房的外部环境、发热量和安装条件后，加装了智能热管设备(HPE)这种间接换热器，以通过空调和热管设备的联动控制来实现机房的解决方案。采用热管技术不需要机械制冷，室内外温差基本保持在6度左右，因而可以适用于全年90％以上的时间。同时，它的部件耗能远远低于传统的压缩机空调，耗能仅为原空调系统的约1/5，因此可以大幅度节省空调耗电。
通常情况下，无论是机房的新建还是扩容，都会充分考虑机房的应用环境，以便选择合适的换热设备。通过热管和热交换器等间接换热设备，可以实现内外环境的隔离，适用范围较广，但是初始投资成本较高。在空气质量较好(没有盐雾、腐蚀性气体污染)、温度和湿度较低、用户有较强的定期维护能力的情况下，可以选择新风系统作为直接换热设备。因此，本公开的另一个应用场景是：采用新风系统和空调进行联动制冷的基站机房。
根据本公开提供的制冷设备控制方案，可以基于大数据技术和神经网络技术，充分考虑到当前室内外温湿度、系统负载等数据，结合负载预测、天气预报、同期历史样本数据等，通过神经网络的计算，提前预测制冷设备负荷、室内温度，并输出制冷设备当日联动控制的最优预案，再与传统控制规则策略相结合，可以实现机房空调和换热设备的可预测的主动控制，达到优化控制、节能降耗的目的。
由于实现了换热设备和空调的可预测的主动联控，显著减少了空调运行时间和启动次数；同时，可将机房设备的工作温度提高到可控的30-40℃的安全范围，进一步降低了制冷设备的能耗。初步估算，相比单空调制冷方式，通过空调和换热设备的主动预测式联控方案，每年可为通信基站节省耗电量将近1万度，平均耗电量减少40％，如果按照500万基站的10％比例计算，每年将减少电费50亿元、以及 135万吨碳排放，经济和社会效益显著。
基于相同的技术构思，本公开还提供一种制冷设备控制装置。
图10和图11为本公开提供的制冷设备控制装置的结构示意图。
如图10所示，本公开提供的制冷设备控制装置包括第一处理模块101、第二处理模块102和控制模块103。
第一处理模块101用于确定当前室外温度。
第二处理模块102用于：将制冷设备负荷的同期历史样本数据和预设的影响因子作为第一输入参数输入第一神经网络模型，得到制冷设备当日预测的负荷；将室外温度的同期历史样本数据和所述制冷设备当日预测的负荷作为第二输入参数输入第二神经网络，得到当日预测的室内温度；以及将所述当日预测的室内温度和预设的制冷效率因子作为第三输入参数输入第三神经网络，得到所述制冷设备当日的最优控制参数。
控制模块103用于根据所述最优控制参数控制所述制冷设备运行。
如图11所示，本公开的制冷设备控制装置还可以包括模型建立模块104。
模型建立模块104用于在初始化阶段建立所述第一神经网络模型、第二神经网络模型和第三神经网络模型。模型建立模块104可以用于：获取历史样本数据，所述样本数据包括室外温度、室内温度和制冷设备负荷；对所述历史样本数据仿真模拟，计算得到制冷设备每日的最优控制参数；以及根据所述历史样本数据和所述制冷设备每日的最优控制参数，建立第一神经网络模型、第二神经网络模型和第三神经网络模型。
模型建立模块104还可以用于：在对所述历史样本数据仿真模拟，计算得到制冷设备每日的最优控制参数之后，并且在根据所述历史样本数据和所述制冷设备每日的最优控制参数，建立第一神经网络模型、第二神经网络模型和第三神经网络模型之前，对所述历史样本数据和所述制冷设备每日的最优控制参数进行归一化处理；以及根据归一化处理后的数据建立训练样本数据集，所述训练样本数据集包括 训练集、验证集和测试集。
模型建立模块104可以用于根据所述训练样本数据集建立第一神经网络模型、第二神经网络模型和第三神经网络模型。
模型建立模块104可以用于：将制冷设备负荷的同期历史样本数据和所述预设的影响因子作为第一输入参数，并将所述制冷设备当日负荷的历史样本数据作为第一输出参数，建立第一神经网络模型；将室外温度的同期历史样本数据和所述制冷设备当日负荷的历史样本数据作为第二输入参数，并将当日室内温度的历史样本数据作为第二输出参数，建立第二神经网络模型；以及将所述当日室内温度的历史样本数据和所述预设的制冷效率因子作为第三输入参数，并将制冷设备当日最优控制参数的历史样本数据作为第三输出参数，建立第三神经网络模型。
样本数据可以包括模拟数据和采样数据。模拟数据是当室内温度大于预设的第三阈值时，经过模拟制冷设备的运行得到的数据。采样数据是当室内温度小于预设的第六阈值且制冷设备实际停机时长大于预设的第七阈值时采样得到的数据。
第一处理模块101可以用于，确定当前时刻前预设时长内的室外温度；以及根据所述当前时刻前预设时长内的室外温度、当天预测温度和预设的第一权重和第二权重确定当前室外温度。
最优控制参数可以包括开启时刻和开启时长。影响因子可以包括以下之一或任意组合：节假日影响因子、潮汐影响因子、区域事件因子。
控制模块103可以用于：若当前室内温度小于或等于预设的第一阈值且大于或等于预设的第二阈值，且满足第一高温预启动条件，则将空调最大运行时长设置为空调开启时长和预设的空调最大开启时长中的最小值，并启动空调，所述第二阈值小于所述第一阈值；以及若空调实际开启时长大于或等于所述空调最大运行时长，则关闭空调。
第一高温预启动条件可以包括：到达所述空调开启时刻，且当前室内温度大于预设的第三阈值，且空调实际停机时长大于预设的空 调最短停机时长。
控制模块103还可以用于：在根据所述最优控制参数控制所述制冷设备运行过程中，若当前室内温度大于所述第一阈值且空调实际停机时长大于所述空调最短停机时长，则将空调最大运行时长设置为所述空调最大开启时长，并启动空调；和/或，若当前室内温度小于所述第二阈值，则关闭空调。
控制模块103还可以用于：若满足第二高温预启动条件，则启动换热设备；以及若所述换热设备实际开启时长大于或等于所述换热设备开启时长，则关闭所述换热设备。
当所述换热设备为间接换热设备时，第二高温预启动条件可以包括：到达所述换热设备开启时刻，且当前室内温度大于预设的第四阈值，且当前室内温度与室外温度的差值大于预设的第五阈值。
当所述换热设备为直接换热设备时，第二高温预启动条件可以包括以下之一：到达所述换热设备开启时刻，且当前室内温度大于预设的第四阈值，且当前室内温度与室外温度的差值大于预设的第八阈值，所述第八阈值大于所述第五阈值；以及到达所述换热设备开启时刻，且当前室内温度大于预设的第四阈值，且当前室内温度与室外温度的差值大于预设的第八阈值，且当前室内湿度小于或等于预设的第九阈值。
当所述换热设备为直接换热设备时，所述空调开启时刻与所述换热设备开启时刻不同；控制模块103还可以用于：若开启所述空调，则关闭所述换热设备；若开启所述换热设备，则关闭所述空调。
控制模块103还可以用于：在根据所述最优控制参数控制所述制冷设备运行之后，若空调当日的实际运行参数与空调当日的最优控制参数之间的误差超过预设的第十阈值，则指示第二处理模块102再次确定空调当日的最优控制参数；以及根据再次确定的空调当日的最优控制参数更新所述训练样本数据集。
控制模块103还可以用于：若当前运行的一种制冷设备故障且另一种制冷设备正常，则关闭所述故障的制冷设备，并开启所述正常的制冷设备；以及若当前运行的两种制冷设备均故障，则在故障消除 时，启动故障消除的制冷设备。
第二处理模块102还可以用于：若当前室内温度小于预设的第六阈值且所述制冷设备实际停机时长大于预设的第七阈值，则根据当前获取的样本数据训练所述第二神经网络模型，所述样本数据包括室外温度、室内温度和制冷设备负荷。
本公开还提供一种计算机设备，该计算机设备包括一个或多个处理器以及存储装置，存储装置上存储有一个或多个程序，当所述一个或多个程序被所述一个或多个处理器执行时，使得所述一个或多个处理器实现本公开提供的制冷设备控制方法。
本公开还提供一种计算机可读介质，其上存储有计算机程序，该计算机程序被处理器执行时，使得所述处理器实现本公开提供的制冷设备控制方法。
本领域普通技术人员可以理解，上文中所公开方法中的全部或某些步骤、装置中的功能模块/单元可以被实施为软件、固件、硬件及其适当的组合。在硬件实施方式中，在以上描述中提及的功能模块/单元之间的划分不一定对应于物理组件的划分；例如，一个物理组件可以具有多个功能，或者一个功能或步骤可以由若干物理组件合作执行。某些物理组件或所有物理组件可以被实施为由处理器，如中央处理器、数字信号处理器或微处理器执行的软件，或者被实施为硬件，或者被实施为集成电路，如专用集成电路。这样的软件可以分布在计算机可读介质上，计算机可读介质可以包括计算机存储介质(或非暂时性介质)和通信介质(或暂时性介质)。如本领域普通技术人员公知的，术语计算机存储介质包括在用于存储信息(诸如计算机可读指令、数据结构、程序模块或其他数据)的任何方法或技术中实施的易失性和非易失性、可移除和不可移除介质。计算机存储介质包括但不限于RAM、ROM、EEPROM、闪存或其他存储器技术、CD-ROM、数字多功能盘(DVD)或其他光盘存储、磁盒、磁带、磁盘存储或其他磁存储装置、或者可以用于存储期望的信息并且可以被计算机访问的任何其他的介质。此外，本领域普通技术人员公知的是，通信介质通常包含计算机可读指令、数据结构、程序模块或者诸如载波或其他传输机 制之类的调制数据信号中的其他数据，并且可包括任何信息递送介质。
本文已经公开了示例实施例，并且虽然采用了具体术语，但它们仅用于并仅应当被解释为一般说明性含义，并且不用于限制的目的。在一些实例中，对本领域技术人员显而易见的是，除非另外明确指出，否则可单独使用与特定实施例相结合描述的特征、特性和/或元素，或可与其他实施例相结合描述的特征、特性和/或元件组合使用。因此，本领域技术人员将理解，在不脱离由所附的权利要求阐明的本公开的范围的情况下，可进行各种形式和细节上的改变。

Claims (19)

  1. 一种制冷设备控制方法，包括：
    确定当前室外温度；
    将制冷设备负荷的同期历史样本数据和预设的影响因子作为第一输入参数输入第一神经网络模型，得到制冷设备当日预测的负荷；
    将室外温度的同期历史样本数据和所述制冷设备当日预测的负荷作为第二输入参数输入第二神经网络，得到当日预测的室内温度；
    将所述当日预测的室内温度和预设的制冷效率因子作为第三输入参数输入第三神经网络，得到所述制冷设备当日的最优控制参数；以及
    根据所述最优控制参数控制所述制冷设备运行。
  2. 如权利要求1所述的方法，其中，在确定当前室外温度的步骤之前，所述方法还包括：
    获取历史样本数据，所述样本数据包括室外温度、室内温度和制冷设备负荷；
    对所述历史样本数据仿真模拟，计算得到制冷设备每日的最优控制参数；以及
    根据所述历史样本数据和所述制冷设备每日的最优控制参数，建立所述第一神经网络模型、第二神经网络模型和第三神经网络模型。
  3. 如权利要求2所述的方法，其中，在对所述历史样本数据仿真模拟，计算得到制冷设备每日的最优控制参数的步骤之后，并且在根据所述历史样本数据和所述制冷设备每日的最优控制参数，建立所述第一神经网络模型、第二神经网络模型和第三神经网络模型的步骤之前，所述方法还包括：
    对所述历史样本数据和所述制冷设备每日的最优控制参数进行归一化处理；以及
    根据归一化处理后的数据建立训练样本数据集，所述训练样本 数据集包括训练集、验证集和测试集，
    根据所述历史样本数据和所述制冷设备每日的最优控制参数，建立所述第一神经网络模型、第二神经网络模型和第三神经网络模型的步骤包括：
    根据所述训练样本数据集建立第一神经网络模型、第二神经网络模型、第三神经网络模型。
  4. 如权利要求2所述的方法，其中，根据所述历史样本数据和所述制冷设备每日的最优控制参数，建立所述第一神经网络模型、第二神经网络模型和第三神经网络模型的步骤包括：
    将制冷设备负荷的同期历史样本数据和所述预设的影响因子作为第一输入参数，并将所述制冷设备当日负荷的历史样本数据作为第一输出参数，建立第一神经网络模型；
    将室外温度的同期历史样本数据和所述制冷设备当日负荷的历史样本数据作为第二输入参数，并将当日室内温度的历史样本数据作为第二输出参数，建立第二神经网络模型；以及
    将所述当日室内温度的历史样本数据和所述预设的制冷效率因子作为第三输入参数，并将制冷设备当日最优控制参数的历史样本数据作为第三输出参数，建立第三神经网络模型。
  5. 如权利要求2所述的方法，其中，所述样本数据包括模拟数据和采样数据，
    所述模拟数据是当室内温度大于预设的第三阈值时，经过模拟制冷设备的运行得到的数据，
    所述采样数据是当室内温度小于预设的第六阈值且所述制冷设备实际停机时长大于预设的第七阈值时采样得到的数据。
  6. 如权利要求1所述的方法，其中，确定当前室外温度的步骤包括：
    确定当前时刻前预设时长内的室外温度；以及
    根据所述当前时刻前预设时长内的室外温度、当天预测温度和预设的第一权重和第二权重确定当前室外温度。
  7. 如权利要求1至6中任一项所述的方法，其中，所述制冷设备包括空调和换热设备，所述最优控制参数包括空调开启时刻、空调开启时长、换热设备开启时刻和换热设备开启时长，所述影响因子包括以下之一或任意组合：节假日影响因子、潮汐影响因子、区域事件因子。
  8. 如权利要求7所述的方法，其中，根据所述最优控制参数控制所述制冷设备运行的步骤包括：
    响应于当前室内温度小于或等于预设的第一阈值且大于或等于预设的第二阈值，并且满足第一高温预启动条件，将空调最大运行时长设置为所述空调开启时长和预设的空调最大开启时长中的最小值，并启动所述空调，所述第二阈值小于所述第一阈值；以及
    响应于空调实际开启时长大于或等于所述空调最大运行时长，关闭所述空调。
  9. 如权利要求8所述的方法，其中，所述第一高温预启动条件包括：
    到达所述空调开启时刻，且当前室内温度大于预设的第三阈值，且空调实际停机时长大于预设的空调最短停机时长。
  10. 如权利要求8所述的方法，其中，在根据所述最优控制参数控制所述制冷设备运行过程中，所述方法还包括：
    响应于当前室内温度大于所述第一阈值且空调实际停机时长大于所述空调最短停机时长，将所述空调最大运行时长设置为所述空调最大开启时长，并启动所述空调；和/或
    响应于当前室内温度小于所述第二阈值，关闭所述空调。
  11. 如权利要求7所述的方法，其中，根据所述最优控制参数控制所述制冷设备运行的步骤包括：
    响应于满足第二高温预启动条件，启动所述换热设备；以及
    响应于换热设备实际开启时长大于或等于所述换热设备开启时长，关闭所述换热设备。
  12. 如权利要求11所述的方法，其中，所述换热设备为间接换热设备，所述第二高温预启动条件包括：
    到达所述换热设备开启时刻，且当前室内温度大于预设的第四阈值，且当前室内温度与室外温度的差值大于预设的第五阈值；或者
    所述换热设备为直接换热设备，所述第二高温预启动条件包括以下之一：
    到达所述换热设备开启时刻，且当前室内温度大于预设的第四阈值，且当前室内温度与室外温度的差值大于预设的第八阈值，所述第八阈值大于所述第五阈值；以及
    到达所述换热设备开启时刻，且当前室内温度大于预设的第四阈值，且当前室内温度与室外温度的差值大于预设的第八阈值，且当前室内湿度小于或等于预设的第九阈值。
  13. 如权利要求7所述的方法，其中，所述换热设备为直接换热设备，所述空调开启时刻与所述换热设备开启时刻不同，并且所述方法还包括：
    响应于开启所述空调，关闭所述换热设备；以及
    响应于开启所述换热设备，关闭所述空调。
  14. 如权利要求3所述的方法，其中，所述制冷设备包括空调，并且在根据所述最优控制参数控制所述制冷设备运行的步骤之后，所述方法还包括：
    响应于所述空调当日的实际运行参数与所述空调当日的最优控制参数之间的误差超过预设的第十阈值，再次确定所述空调当日的最 优控制参数；以及
    根据再次确定的所述空调当日的最优控制参数更新训练样本数据集。
  15. 如权利要求1所述的方法，还包括：
    响应于当前运行的一种制冷设备故障且另一种制冷设备正常，关闭所述故障的制冷设备，并开启所述正常的制冷设备；以及
    响应于当前运行的两种制冷设备均故障，在故障消除时，启动故障消除的制冷设备。
  16. 如权利要求1所述的方法，其中，在根据所述最优控制参数控制所述制冷设备运行的过程中，所述方法还包括：
    响应于当前室内温度小于预设的第六阈值且所述制冷设备实际停机时长大于预设的第七阈值，根据当前获取的样本数据训练所述第二神经网络模型，所述样本数据包括室外温度、室内温度和制冷设备负荷。
  17. 一种制冷设备控制装置，包括：第一处理模块、第二处理模块和控制模块，
    所述第一处理模块用于确定当前室外温度；
    所述第二处理模块用于：
    将制冷设备负荷的同期历史样本数据和预设的影响因子作为第一输入参数输入第一神经网络模型，得到制冷设备当日预测的负荷；
    将室外温度的同期历史样本数据和所述制冷设备当日预测的负荷作为第二输入参数输入第二神经网络，得到当日预测的室内温度；以及
    将所述当日预测的室内温度和预设的制冷效率因子作为第三输入参数输入第三神经网络，得到所述制冷设备当日的最优控制参数；
    所述控制模块用于根据所述最优控制参数控制所述制冷设备运行。
  18. 一种计算机设备，包括：
    一个或多个处理器；
    存储装置，其上存储有一个或多个程序；
    当所述一个或多个程序被所述一个或多个处理器执行时，使得所述一个或多个处理器实现如权利要求1至16中任一项所述的制冷设备控制方法。
  19. 一种计算机可读介质，其上存储有计算机程序，其中，所述程序被处理器执行时，使得所述处理器实现如权利要求1至16中任一项所述的制冷设备控制方法。
PCT/CN2021/099313 2020-06-10 2021-06-10 制冷设备控制方法、装置、计算机设备和计算机可读介质 Ceased WO2021249461A1 (zh)

Priority Applications (3)

Application Number Priority Date Filing Date Title
JP2022576079A JP7473690B2 (ja) 2020-06-10 2021-06-10 冷却機器の制御方法、冷却機器制御装置、コンピュータ機器及びコンピュータ可読媒体
EP21821753.7A EP4166862A4 (en) 2020-06-10 2021-06-10 METHOD AND APPARATUS FOR CONTROLLING A COOLING DEVICE, COMPUTER DEVICE AND COMPUTER-READABLE MEDIUM
BR112022025218A BR112022025218A2 (pt) 2020-06-10 2021-06-10 Método e dispositivo de controle de equipamento de refrigeração, dispositivo computacional e meio legível por computador

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202010523728.8A CN113776171B (zh) 2020-06-10 2020-06-10 制冷设备控制方法、装置、计算机设备和计算机可读介质
CN202010523728.8 2020-06-10

Publications (1)

Publication Number Publication Date
WO2021249461A1 true WO2021249461A1 (zh) 2021-12-16

Family

ID=78834644

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/099313 Ceased WO2021249461A1 (zh) 2020-06-10 2021-06-10 制冷设备控制方法、装置、计算机设备和计算机可读介质

Country Status (5)

Country Link
EP (1) EP4166862A4 (zh)
JP (1) JP7473690B2 (zh)
CN (1) CN113776171B (zh)
BR (1) BR112022025218A2 (zh)
WO (1) WO2021249461A1 (zh)

Cited By (16)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114593474A (zh) * 2022-03-11 2022-06-07 广东美的暖通设备有限公司 一种喷淋控制方法、装置、电子设备及存储介质
CN114626616A (zh) * 2022-03-21 2022-06-14 特斯联科技集团有限公司 园区高温尾气余热回收cchp的最优运行方法及系统
CN115076878A (zh) * 2022-05-30 2022-09-20 青岛海尔空调器有限总公司 空调外机自清洁方法、装置、计算机可读存储介质及空调
CN115164361A (zh) * 2022-06-13 2022-10-11 中国电信股份有限公司 一种数据中心控制方法、装置、电子设备和存储介质
CN115789911A (zh) * 2022-11-17 2023-03-14 中国联合网络通信集团有限公司 一种空调控制方法、装置、电子设备及存储介质
CN116017935A (zh) * 2022-12-06 2023-04-25 北京纪新泰富机电技术股份有限公司 机房控制设备运行参数调整方法及装置、设备及存储介质
CN116147154A (zh) * 2022-11-01 2023-05-23 中国电信股份有限公司 机房空调温度调节方法、装置，及电子设备
CN116292244A (zh) * 2023-02-23 2023-06-23 中车唐山机车车辆有限公司 列车空压机的性能检测方法、列车及存储介质
CN116608551A (zh) * 2023-05-19 2023-08-18 华润数字科技有限公司 冷负荷预测方法和装置、电子设备及存储介质
WO2023236553A1 (zh) * 2022-06-10 2023-12-14 青岛海尔空调器有限总公司 空调系统控制方法、装置、电子设备、存储介质及产品
CN117234080A (zh) * 2023-09-21 2023-12-15 江苏亚奥科技股份有限公司 一种面向大型动环监控场景的室温智能调控方法
CN117704566A (zh) * 2024-01-15 2024-03-15 南京龟兔赛跑软件研究院有限公司 基于新能源风柜数据识别的室内温湿度控制系统及方法
CN117970986A (zh) * 2024-04-01 2024-05-03 广东热矩智能科技有限公司 一种冷热系统的温湿度控制方法、装置及介质
CN118192300A (zh) * 2024-04-10 2024-06-14 中建安装集团有限公司 一种基于仿真平台的能源数据管理系统及方法
CN120872624A (zh) * 2025-09-29 2025-10-31 浪潮计算机科技有限公司 运行参数的调整方法、装置、电子设备及存储介质
CN121576753A (zh) * 2026-01-22 2026-02-27 杭州康钡电机有限公司 一种冷柜温度控制系统、方法和冷柜

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113847715B (zh) * 2020-06-28 2024-01-02 中兴通讯股份有限公司 基站的空调调控的方法以及装置、电子设备、介质
CN114222477B (zh) * 2021-12-13 2024-09-03 中国联合网络通信集团有限公司 数据中心的节能控制方法、装置、存储介质及程序产品
CN116857798A (zh) * 2023-06-25 2023-10-10 珠海格力电器股份有限公司 空调用电预测方法、电网负荷预测方法及其装置
CN117469774B (zh) * 2023-12-28 2024-04-02 北京市农林科学院智能装备技术研究中心 空调系统调控方法、装置、电子设备及存储介质
CN120292701B (zh) * 2025-05-12 2025-12-16 和欣智能科技河北雄安有限公司 基于智能化的空调资源节能控制方法及系统

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH08210689A (ja) * 1995-02-07 1996-08-20 Hitachi Plant Eng & Constr Co Ltd 空調熱負荷予測システム
WO2009064111A2 (en) * 2007-11-12 2009-05-22 The Industry & Academic Cooperation In Chungnam National University Method for predicting cooling load
CN105928292A (zh) * 2016-04-20 2016-09-07 山东三九制冷设备有限公司 一种基于神经网络的负荷预测与需求响应控制的光伏冷库系统
CN109130767A (zh) * 2017-06-28 2019-01-04 北京交通大学 基于客流的轨道交通车站通风空调系统的智能控制方法
CN109871987A (zh) * 2019-01-28 2019-06-11 中建八局第三建设有限公司 一种智能建筑暖通设备综合节能控制方法
CN109945420A (zh) * 2019-03-26 2019-06-28 南京南瑞继保电气有限公司 基于负荷预测的空调控制方法、装置及计算机存储介质
KR20200052437A (ko) * 2018-10-29 2020-05-15 에스케이텔레콤 주식회사 소규모 이력 데이터에서 냉방부하를 예측하기 위한 학습 방법 및 장치

Family Cites Families (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2704026B2 (ja) * 1990-05-08 1998-01-26 株式会社東芝 熱負荷予測システム
AU2010362490B2 (en) * 2010-10-13 2015-09-03 Weldtech Technology (Shanghai) Co., Ltd. Energy-saving optimized control system and method for refrigeration plant room
US10353355B2 (en) 2015-05-18 2019-07-16 Mitsubishi Electric Corporation Indoor environment model creation device
JP6807556B2 (ja) * 2015-10-01 2021-01-06 パナソニックＩｐマネジメント株式会社 空調制御方法、空調制御装置及び空調制御プログラム
JP6702376B2 (ja) * 2018-09-03 2020-06-03 ダイキン工業株式会社 送風制御装置
CN109341010B (zh) * 2018-09-19 2021-04-30 新智能源系统控制有限责任公司 一种电制冷机空调系统用供能一体化的控制方法和装置
CN109323425B (zh) * 2018-11-15 2021-05-25 广东美的制冷设备有限公司 空调的控制方法、装置及可读存储介质
CN110186156A (zh) * 2019-06-03 2019-08-30 西安锦威电子科技有限公司 制冷站模糊控制系统
CN110410942B (zh) * 2019-07-30 2020-12-29 上海朗绿建筑科技股份有限公司 一种冷热源机房节能优化控制方法及系统
CN110864414B (zh) * 2019-10-30 2021-09-24 郑州电力高等专科学校 基于大数据分析的空调用电负荷智能控制调度方法

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH08210689A (ja) * 1995-02-07 1996-08-20 Hitachi Plant Eng & Constr Co Ltd 空調熱負荷予測システム
WO2009064111A2 (en) * 2007-11-12 2009-05-22 The Industry & Academic Cooperation In Chungnam National University Method for predicting cooling load
CN105928292A (zh) * 2016-04-20 2016-09-07 山东三九制冷设备有限公司 一种基于神经网络的负荷预测与需求响应控制的光伏冷库系统
CN109130767A (zh) * 2017-06-28 2019-01-04 北京交通大学 基于客流的轨道交通车站通风空调系统的智能控制方法
KR20200052437A (ko) * 2018-10-29 2020-05-15 에스케이텔레콤 주식회사 소규모 이력 데이터에서 냉방부하를 예측하기 위한 학습 방법 및 장치
CN109871987A (zh) * 2019-01-28 2019-06-11 中建八局第三建设有限公司 一种智能建筑暖通设备综合节能控制方法
CN109945420A (zh) * 2019-03-26 2019-06-28 南京南瑞继保电气有限公司 基于负荷预测的空调控制方法、装置及计算机存储介质

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP4166862A4 *

Cited By (22)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114593474B (zh) * 2022-03-11 2023-12-19 广东美的暖通设备有限公司 一种喷淋控制方法、装置、电子设备及存储介质
CN114593474A (zh) * 2022-03-11 2022-06-07 广东美的暖通设备有限公司 一种喷淋控制方法、装置、电子设备及存储介质
CN114626616A (zh) * 2022-03-21 2022-06-14 特斯联科技集团有限公司 园区高温尾气余热回收cchp的最优运行方法及系统
CN114626616B (zh) * 2022-03-21 2023-11-21 特斯联科技集团有限公司 园区高温尾气余热回收cchp的最优运行方法及系统
CN115076878A (zh) * 2022-05-30 2022-09-20 青岛海尔空调器有限总公司 空调外机自清洁方法、装置、计算机可读存储介质及空调
WO2023236553A1 (zh) * 2022-06-10 2023-12-14 青岛海尔空调器有限总公司 空调系统控制方法、装置、电子设备、存储介质及产品
CN115164361A (zh) * 2022-06-13 2022-10-11 中国电信股份有限公司 一种数据中心控制方法、装置、电子设备和存储介质
CN115164361B (zh) * 2022-06-13 2024-06-07 中国电信股份有限公司 一种数据中心控制方法、装置、电子设备和存储介质
CN116147154A (zh) * 2022-11-01 2023-05-23 中国电信股份有限公司 机房空调温度调节方法、装置，及电子设备
CN115789911A (zh) * 2022-11-17 2023-03-14 中国联合网络通信集团有限公司 一种空调控制方法、装置、电子设备及存储介质
CN115789911B (zh) * 2022-11-17 2024-05-03 中国联合网络通信集团有限公司 一种空调控制方法、装置、电子设备及存储介质
CN116017935A (zh) * 2022-12-06 2023-04-25 北京纪新泰富机电技术股份有限公司 机房控制设备运行参数调整方法及装置、设备及存储介质
CN116292244A (zh) * 2023-02-23 2023-06-23 中车唐山机车车辆有限公司 列车空压机的性能检测方法、列车及存储介质
CN116292244B (zh) * 2023-02-23 2025-09-23 中车唐山机车车辆有限公司 列车空压机的性能检测方法、列车及存储介质
CN116608551A (zh) * 2023-05-19 2023-08-18 华润数字科技有限公司 冷负荷预测方法和装置、电子设备及存储介质
CN117234080A (zh) * 2023-09-21 2023-12-15 江苏亚奥科技股份有限公司 一种面向大型动环监控场景的室温智能调控方法
CN117234080B (zh) * 2023-09-21 2024-05-28 江苏亚奥科技股份有限公司 一种面向大型动环监控场景的室温智能调控方法
CN117704566A (zh) * 2024-01-15 2024-03-15 南京龟兔赛跑软件研究院有限公司 基于新能源风柜数据识别的室内温湿度控制系统及方法
CN117970986A (zh) * 2024-04-01 2024-05-03 广东热矩智能科技有限公司 一种冷热系统的温湿度控制方法、装置及介质
CN118192300A (zh) * 2024-04-10 2024-06-14 中建安装集团有限公司 一种基于仿真平台的能源数据管理系统及方法
CN120872624A (zh) * 2025-09-29 2025-10-31 浪潮计算机科技有限公司 运行参数的调整方法、装置、电子设备及存储介质
CN121576753A (zh) * 2026-01-22 2026-02-27 杭州康钡电机有限公司 一种冷柜温度控制系统、方法和冷柜

Also Published As

Publication number Publication date
CN113776171A (zh) 2021-12-10
BR112022025218A2 (pt) 2023-01-03
CN113776171B (zh) 2024-02-13
EP4166862A1 (en) 2023-04-19
JP7473690B2 (ja) 2024-04-23
EP4166862A4 (en) 2024-06-26
JP2023530628A (ja) 2023-07-19

Similar Documents

Publication Publication Date Title
WO2021249461A1 (zh) 制冷设备控制方法、装置、计算机设备和计算机可读介质
CN201765486U (zh) 基于云计算的设备监控系统
US9350562B2 (en) Energy management control system and method based on cloud computing
CN120295260A (zh) 一种智能楼宇节能减排数字孪生管理系统及方法
US9982903B1 (en) HVAC system with predictive free cooling control based on the cost of transitioning into a free cooling state
CN102844773B (zh) 基于交互式学习的建筑物节能化单元和系统
CN201812187U (zh) 基于云计算的电子信息系统机房能源管理控制系统
WO2011106914A1 (zh) 基于云计算的设备监控系统及方法
CN108168052B (zh) 一种中央空调制冷系统最优启停控制方法
WO2021208313A1 (zh) 一种供热、供电、制冷一体自然能源智慧系统及控制方法
CN109059195B (zh) 用于削减电网负荷峰值的中央空调的控制方法及控制系统
WO2011106918A1 (zh) 基于云计算的电子信息系统机房能源管理控制系统及方法
CN104238531B (zh) 一种铁路车站能源管理系统和节能控制方法
CN116085936A (zh) 一种中央空调能源站智慧能源管理系统、设备及介质
CN102980276B (zh) 一种基站节能用智能新风系统
CN103064389B (zh) 一种智能能耗管理支撑系统
US20230236560A1 (en) Net zero energy facilities
KR20240160164A (ko) 건물의 hvac 컴포넌트의 동적 제어를 위한 시스템 및 방법
CN108895633A (zh) 利用建筑结构作为蓄冷介质的中央空调系统控制方法
CN113418228A (zh) 基于供需匹配的空气源热泵变回差水温控制方法及系统
US20250231540A1 (en) Building control system with net zero energy consumption and carbon emissions
CN120650835A (zh) 中央空调能耗优化控制方法、系统、电子设备和存储介质
CN111928428B (zh) 一种考虑需求响应的空调系统的控制方法及制冷系统
TW201027014A (en) Method for managing air conditioning power consumption
CN120969918B (zh) 基于动态负荷的光伏光热热泵供暖控制方法

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21821753

Country of ref document: EP

Kind code of ref document: A1

ENP Entry into the national phase

Ref document number: 2022576079

Country of ref document: JP

Kind code of ref document: A

REG Reference to national code

Ref country code: BR

Ref legal event code: B01A

Ref document number: 112022025218

Country of ref document: BR

ENP Entry into the national phase

Ref document number: 112022025218

Country of ref document: BR

Kind code of ref document: A2

Effective date: 20221209

NENP Non-entry into the national phase

Ref country code: DE

ENP Entry into the national phase

Ref document number: 2021821753

Country of ref document: EP

Effective date: 20230110