JPH096407A - Dynamic characteristic model parameter tuning support device - Google Patents

Abstract

(57) [Summary] [Structure] Simplified model calculation means 1 for expressing the device element by the transfer function model 7 using the parameters of the detailed dynamic characteristic model, and approximating the input and output of the element block by the transfer function model 1, Actual waveform characteristic extracting means 2 for calculating a value 11 such as a gain or time constant of a transfer function from a step response waveform 10 of an actual machine and a value 16 such as a gain or time constant of a transfer function used in the means 1 and detailed dynamic characteristics for giving them Parameter of model and its value and relational expression 17, value 11 such as gain and time constant of transfer function calculated by means 2 and modification of parameter for matching simplified model transfer function 7 and transfer function calculated by means 2 It comprises means 3 for displaying the quantity. [Effect] Tuning of the plant simulator dynamic characteristic model parameter can be accurately performed by a logically clear method based on modeling of a physical phenomenon, and the time required for adjusting the control system of the plant can be shortened.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a device for supporting parameter tuning of a dynamic characteristic model used in a plant simulator such as a thermal power plant.

[0002]

2. Description of the Related Art In order to tune the parameters of a dynamic characteristic model so that the simulation result of a thermal power plant does not greatly differ from that of an actual plant, in the conventional method, a trained expert uses a boiler input such as a feed water flow rate or a fuel flow rate. It is carried out by trial and error by comparing the characteristic quantities (gain and time constant) of the actual data of the step response characteristics and analyzing the sensitivity to adjust the system parameters.

In the case of tuning the control system, the entire plant is regarded as one black box, and a reference model for appropriately assuming the input / output shape is used first, and the response time constant and gain assumed here are used. From the relationship with the controller parameters (generally, the response width reduction ratio is specified), the correction amount is determined and tuning is performed.

In the case of parameter tuning of a model, even in the field of analog electronic circuit design in which there are many parameters to be adjusted, an expert spends a great deal of time while considering a wide variety of characteristics, and direct current characteristics, alternating current characteristics, transient characteristics Although each circuit constant value was decided optimally by trial and error so that the design specification value was satisfied for the circuit characteristics such as
As a method for searching for optimization, a method has been proposed in which circuit constants having high contribution are minutely changed and optimized while learning the effectiveness of contribution to the change of each circuit constant value (JP-A-5-120364). Gazette).

[0005]

At present, even in a thermal power plant, from the viewpoint of stable power supply, even if one of a plurality of auxiliary machines such as a water supply pump stops, the plant is operated without stopping (operation by a runback function). ) The need to let is increased. (Hereinafter, such operation is referred to as special operation in order to distinguish it from normal rated output operation.) In special operation, the limits such as pressure such as 33 important plant variables are set by the boiler protection interlock. Do not exceed the value
It is necessary to adjust the control system for special operation. Since the control system for special operation is roughly set in advance by simulation of the thermal power plant, if the simulation result is significantly different from the actual machine, there is a problem that it takes a long time to adjust the control system. From this,
Providing a device that supports tuning of dynamic characteristic model parameters used in a plant simulator is
This is a very important issue.

In order to simulate a plant system such as a thermal power plant system that is configured as a combination of a plurality of equipment elements and has a small amount of measurement data inside the plant, a physical phenomenon that occurs in each equipment element that constitutes the plant is analyzed for each equipment element. In the plant simulator configured with the detailed dynamic characteristic model modeled in Fig. 1, one parameter is related to the dynamic characteristics of many physical quantities. The method of extracting and changing such parameters may not be able to tune the parameters efficiently.

An object of the present invention is to provide a simulator for a plant system constituted by a combination of a plurality of equipment elements such as a thermal power plant when a dynamic characteristic model representing a physical phenomenon is used. An object is to provide a support device for tuning a parameter of a dynamic characteristic model.

[0008]

In order to achieve the above object, parameter tuning is performed by combining the following means.

(Means 1) The entire plant is divided into actual device data measurement points to form element blocks, and the input and output of the element blocks are transferred by using the parameters of the detailed dynamic characteristic model in each device element in the element blocks. A simplified model calculation means that approximates with a model.

(Means 2) Assuming that the step response waveform of the actual machine can be represented by a model of the same order as the transfer function model,
Actual waveform feature extraction means for calculating, from the step response waveform, a gain, a time constant, etc. of a transfer function corresponding to the device element existing between actual machine data measurement points.

(Means 3) Gains and time constants of the transfer function of the element block obtained by the simplified model calculation means, parameters and values of the detailed dynamic characteristic model giving them, and the actual waveform feature extraction means Means for displaying the gain or time constant value of the transfer function calculated in step 1, and the correction amount of the parameter for matching the simplified model transfer function with the transfer function calculated by the means.

[0012]

A complex plant system model is constructed on the basis of a detailed dynamic characteristic model by means of a simplified model calculating means for approximating the input / output of an element block by a transfer function model, and thereby a complex plant system model is represented by parameters. Therefore, the actual waveform feature extraction means can identify the parameter of the element block model.

Then, the gain and time constant values of the transfer function obtained by the simplified model calculation means, the parameters and their values of the detailed dynamic characteristic model giving them, and the transfer function calculated by the actual waveform feature extraction means Tuning of plant simulator dynamic characteristic model parameters is supported by means for displaying the values of gains and time constants, and the amount of parameter modification for matching the simplified model transfer function with the transfer function calculated by the means.

[0014]

FIG. 1 shows an embodiment of the present invention. In this embodiment, the simplified model calculation means 1, the performance feature extraction means 2, the step response waveform 13 of the actual machine, the means 3 for displaying the information 14 regarding the parameter tuning of the detailed dynamic characteristic model of the plant simulator 4, and the plant simulator 4 are used. A plant simulator dynamic characteristic model parameter tuning support device 20 is configured.

From the detailed dynamic characteristic model 5 mounted on the plant simulator 4, the user 6 creates a transfer function model 7 corresponding to each device element that constitutes the plant, and based on this, the simplified model calculation means 1 is constructed. To do. The apparatus 20 outputs the simplified model response waveform 9 by calculating and inputting the step input data 8 of the plant to the simplified model calculation means 1. Further, by inputting the response waveform data 10 to the step input 8 of the plant to the actual waveform characteristic extraction means 2 for calculating the gain and time constant of the transfer function based on the transfer function model 7 corresponding to each device element, the transfer is performed. The actual waveform characteristic amount 11 such as the gain and time constant of the function is output.

Further, in this embodiment, the simulator 4 is built in, and the simulation result 12 is output by inputting the step input data 8 of the plant into the simulator 4. Input the output results 9, 11 and 12 of the means 1, 2 and 4 to the display means 3,
On the display of the device of the present invention, a group of response curves 13 to step inputs by means 1, 2 and 4 and information 14 on tuning of detailed dynamic characteristic model parameters of the simulator 4 (that is, gain and time of transfer function obtained by means 2). The constant 11, the gain and time constant 16 of the transfer function obtained by the means 1, the mathematical expression 17 indicating the relationship between the time constant or gain of the transfer function model 7 and the dynamic characteristic model parameter, and the time point used for the dynamic characteristic model parameter. Display the numerical value 18) in. When the user 6 tunes the parameters of the detailed dynamic characteristic model 5 of the plant simulator 4, the user 6 refers to these because information 14 regarding the tuning of the detailed dynamic characteristic model parameters is displayed. Formula 1 showing the relationship between constants and gains and dynamic characteristic model parameters
The parameters can be tuned while being based on the numerical range of the dynamic characteristic model parameters constrained by 7. Therefore, the tuning work load is reduced.

FIG. 2 is a block diagram showing the structure of a part of a simplified model of a plant which is the object of the present invention. Here, 21 is an element block showing a steam pipe and is composed of two equipment element steam models 22 and a metal model 23, and 24 is an element block consisting of one equipment element showing a brackish water separator. is there. Inlet flow rate G _in 25, steam pipe heat input Q _Bin 26, outlet flow rate G _out 28, outlet enthalpy H _out 29, and brackish water separator enthalpy H _sep 30,
It is assumed that the brackish water separator pressure P _sep 31 can be measured. Also,
The amount of heating the water in the steam pipe is Q _MFin 27.

The simplified model calculation means 1 creates a transfer function model based on the detailed dynamic characteristic model of the simulator corresponding to the device and uses it in the means 1. For example, device element 2
When the transfer function models of 2,23 are created by approximation from the physical model,

[0019]

[ _Equation 1] G _out = a ₁ G _in + a ₂ Q _MFin + a ₃ P _sep ( _Equation 1)

[0020]

[ _Equation 2] H _out = a ₄ G _in + a ₅ Q _MFin + a ₆ P _sep ( _Equation 2)

[0021]

_Equation 3 Q _MFin = 1 / (1 + sτ _m ) Q _Bin (Equation 3) Here, a _{1 to} a ₆ are terms indicating dynamic characteristics with respect to input, and τ _m is a time constant relating to heat transfer of metal. Further, a ₃ , a ₆ and τ _m are functions of the pressure loss coefficient ξ ₁ and the heat transfer coefficient C _p which are the parameters of the dynamic characteristic model, and the others are constants.

[0022]

[Formula 4] a ₃ = b ₁ / ξ ₁ (Formula 4)

[0023]

[Formula 5] a ₆ = b ₂ / ξ ₁ + b ₃ (Formula 5)

[0024]

[6] _{τ m = b 4 C p ...} ( 6) where, by substituting Expression 3 number 1, the number 2, simplified and approximated to first order lag transfer function of the element block 21 showing the steam pipe response A model for calculating is obtained.

[0025]

[Equation 7] G _out = a ₁ G _in + a ₂ / (1 + sτ _m ) Q _Bin + a ₃ P _sep (Equation 7)

[0026]

[Equation 8] H _out = a ₄ G _in + a ₅ / (1 + sτ _m ) Q _Bin + a ₆ P _sep (Equation 8) Thus, the simplified model calculation means 1 is configured by the model for calculating the simplified response. .

On the other hand, the gain a _{2 of the} transfer function corresponding to the actual machine data, which is obtained by the means 2, is obtained by assuming that the step response waveform of the actual machine is characterized by the same transfer function as that represented by the transfer function model. And time constant τ _m are calculated. Therefore, the means represented by the parameters of the detailed dynamic characteristic model so that the time constant τ _m and the gain a _{2 of} the transfer function approximation model used in the means 1 match the values of the time constant obtained in the means 2. The parameters of the dynamic characteristic model may be tuned while referring to the relational expressions such as the time constant of the transfer function used in 1. That is, by means 3,
Since the relational expressions (obtained from the detailed dynamic characteristic model) such as the time constant of the transfer function used in the means 1 and the values thereof and the values such as the gain and time constant of the transfer function calculated in the means 2 are displayed, these The parameters can be tuned while viewing the numerical information required for tuning.
Further, when values such as the time constant of the transfer function used in the means 1 cannot be physically obtained from the detailed dynamic characteristic model, the means 2
The value is calculated back from the gain and time constant of the transfer function calculated in step 2. If the value is found, it is displayed so that it can be distinguished from other definite values, and if it is not found, a question mark or the like is displayed.

FIG. 3A shows an embodiment relating to the display section of the device of the present invention. The display unit 51 is composed of, for example, a multi-window, and in the window 52, a calculation result 53a obtained by a plant simulator, a step input response curve 53b obtained by an actual plant, and a simplified model calculation means 1 are used. The simplified model response waveform 53c is displayed. Further, in the windows 54a and 54b, the values of the gain and time constant of the transfer function used in the means 1, the parameters of the detailed dynamic characteristic model associated with them, and the values thereof, and the means 2
Display the gain and time constant of the transfer function calculated in.

FIG. 3B shows a table 55 displaying items and numerical values. Table 55 displays information about the time constant τ _m of the transfer function used in the means 1. That is, the mathematical expression 58 indicating the relationship between the actual data value 56 calculated by the means 2 and the transfer function model value 57, the transfer function model value 57 and the detailed dynamic characteristic model parameter of the plant simulator used in the means 1 associated with the actual data value 56. , And Equation 58
The detailed dynamic characteristic model parameters 59 and 60 used for are listed. When the calculation error of the plant simulator is large and the user tunes the parameter of the detailed dynamic characteristic model of the plant simulator, the user can tune the parameter while referring to the constraint condition related to the parameter of the detailed dynamic characteristic model of Equation 58. it can. Therefore, the tuning work load is reduced.

[0030]

According to the present invention, the tuning of plant simulator dynamic characteristic model parameters, which has been carried out by experts by trial and error, is accurately performed by a logically clear method based on modeling of physical phenomena. Therefore, the time required for adjusting the control system of the plant can be shortened.

[Brief description of drawings]

FIG. 1 is a block diagram of a dynamic characteristic model parameter tuning support apparatus according to an embodiment of the present invention.

FIG. 2 is a block diagram showing a configuration of an element block of a plant which is a target of the present invention.

FIG. 3 is an explanatory view showing an embodiment relating to a display portion of the device of the present invention.

[Explanation of symbols]

1 ... Simplified model calculation means, 2 ... Actual feature extraction means, 3 ...
Display means, 4 ... Plant simulator, 5 ... Detailed dynamic characteristic model, 6 ... User, 7 ... Transfer function model, 8 ... Step input data, 9 ... Simplified model response waveform, 10 ... Response waveform data, 11 ... Actual waveform feature amount , 12 ... Simulation result, 13 ... Response curve group, 14 ... Information on parameter tuning, 16 ... Gain and time constant of transfer function, 17 ... Numerical expression, 18 ... Numerical value, 20 ... Tuning support device.

Claims (2)

[Claims] 1. In order to simulate a plant system configured as a combination of a plurality of equipment elements, a detailed dynamic characteristic model in which a physical phenomenon occurring in each equipment element constituting the plant is modeled for each equipment element is formed. For the plant simulator, (1) the entire plant is divided into actual device data measurement points to form element blocks, and the input / output of the element blocks is performed using the parameters of the detailed dynamic characteristic model in each device element in the element blocks. Simplified model calculation means approximating with a transfer function model, (2) A transfer function corresponding to the device element existing between actual device data measurement points, as the step response waveform of the actual device can be represented by a model of the same order as the transfer function model Waveform characteristic extracting means for calculating the gain, time constant, etc. of the step response waveform from the step response waveform, and (3) the simplified model calculator. The value of the gain and time constant of the transfer function of the element block obtained in, and the parameter and its value of the detailed dynamic characteristic model to give them, and the gain and time constant of the transfer function calculated by the actual waveform feature extraction means A parameter of the dynamic characteristic model, which is constituted by three means of a value and a means for displaying the correction amount of the parameter for matching the simplified model transfer function with the transfer function calculated by the means. Tuning support device. 2. The response curve of an actual plant of a plant and the response curve by simulation, the gain of the transfer function and the value of the time constant calculated by the actual waveform feature extraction means, and the simplified model calculation means according to claim 1. A parameter tuning support device for a dynamic characteristic model having a function of simultaneously displaying the value of a parameter of a detailed dynamic characteristic model that constitutes the gain and time constant of a transfer function to be used, and its relational expression.