JPS6285301A - Command support methods and systems - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

[Detailed Description of the Invention] [Field of Application of the Invention] The present invention relates to support for reasoning and judgment, and is particularly applicable to command and control systems that are complex, large-scale, and in which algorithms, knowledge, and how to use them are often unclear. related to command support methods and systems.

[Background of the invention]

Although Hearsay-II targets speech understanding and does not go beyond the area of signal processing, this system
S) In particular, the main target is decision support for command and control systems, and the targets are completely different. Also, like this system,
Divide knowledge for knowledge control.

There is no consideration given to hierarchizing, evaluating, adding, modifying, or supporting lower-level knowledge with higher-level knowledge, or reflecting the results in later applications.

[Purpose of the invention]

The purpose of the present invention is to efficiently generate effective decision support data.
Moreover, it is an object of the present invention to provide a suitable command/support method and system for an information processing system, a decision support system, etc. that can be provided even in unexpected situations.

[Summary of the invention]

In order to achieve the above objective, the present invention divides specialized knowledge, particularly into knowledge directly related to a domain, knowledge to effectively utilize these, and knowledge to evaluate, integrate, and add to this knowledge. It is characterized by the provision of an independent reasoning mechanism for interpreting and executing the knowledge and a man-machine interface mechanism for utilizing human knowledge for one or more purposes.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

FIG. 1 is an overall block diagram of an information processing device W1 (judgment support device), particularly a command/control support device, showing an embodiment of the present invention.

In FIG. 1, 1 is a computer, 2 is a display device, 3 is a keyboard, and 4 is an input device called a mouse.

When operating this device, the procedure is first stored in the firmware storage device [11] of the computer 1, or loaded into the main storage device 12 of the computer 1 from an auxiliary storage device [13] such as a disk or magnetic tape, or partially loaded into the main storage device 12 of the computer 1. is stored in the refresh armware, the rest is stored in the loading process, and the computer is activated by controlling the operation of the computer. A computer modified so that its operation follows this method will be referred to as system 1.

FIG. 2 is a functional configuration diagram of the system 1 of the embodiment. This embodiment deals with a command and control system for train schedule organization and schedule rescheduling. The functions of the system 1 include a knowledge unit 15 for storing knowledge regarding the object, an inference unit 16 for performing inference using the knowledge, and an interface with man-machines, sensors, process control computers, etc. It consists of an interface section 17 for

The knowledge section 15 has knowledge necessary for command and control in railways divided into the following formats.

(1) Individual passenger service requests and the creation, evaluation, and
Passenger request button for know-how and regulations for correction! 1510゜(2) Train schedule knowledge 1520, including know-how and regulations for creating, evaluating, and modifying individual train schedules and meeting passenger requirements 1 to i within the constraints of resources such as rolling stock.゜(3) Individual vehicle schedules and train schedules 1-j
Vehicle schedule knowledge 1530° consisting of know-how and regulations for creating, evaluating, and modifying to satisfy Passenger command expertise 1515° to manage the creation and modification of passenger services so that the train schedule as a whole satisfies passenger demands and satisfies resource constraints 1.
Passenger Directive 1515 and Train Directive 153 have been implemented to ensure that there are no contradictions or imbalances between individual train schedules and to ensure efficiency.
Train operation command expertise 1525° to manage the creation and modification of individual train schedules while communicating with 5. (6) Overall check of vehicle allocation, ensuring that the train schedule is satisfied and that each vehicle is Train command specialist knowledge Fa 1535° (7) To manage the creation and modification of individual vehicle schedules so that the loading/unloading and inspection schedules are reasonable, have margins, and are balanced. Which part to use? If there are several alternatives, which one to consider first? Have all or some of the same or similar cases already been inferred? These past inference results can be used as is. knowledge for inferring whether it is effective in determining the consideration order for selecting alternatives.

Furthermore, should the inference results be saved, and if so, in what form (for example, as independent knowledge)?

or as part of other knowledge), how and when to evaluate the knowledge used, and what knowledge to retain for evaluation.

etc., the system Ill control button Ill 540°154
5.1550,1555. Or strategic knowledge 1560,
It is stored in the knowledge section 15 as 1565.

(7-a) The lower control knowledge 1540.1545 is only for controlling the passenger command expert knowledge 1515, or.

Wholesale specializing in train commands! ! This is knowledge that controls knowledge that narrows the scope to one specialized function, such as controlling only a 1 s 3 s. For example, decide which of several pieces of knowledge about train overtaking should be applied first and how to evaluate it.

(7-b) Upper control knowledge 1550.1555 is.

Control knowledge that selects and evaluates alternative knowledge that crosses the specialized knowledge of each directive, lower-level control knowledge 1540.1545
It is also control knowledge for.

For example, it is determined which of the lower-level control knowledge to use, which determines how to use the knowledge for overtaking a train, and how to set the conditions and parameters when using the lower-level control knowledge to be used. At this time, unlike lower-level control knowledge, which of all related specialized knowledge and lower-level control knowledge,
Decide how to apply it. For example, when train A's departure time is set to TAL to solve the inconvenience of overtaking trains.

Suppose that a vehicle conflict occurs. At this time, the lower control knowledge can be used to apply the knowledge to change the vehicle assignment of train A by simply setting the bald time of train A to TA2 or setting the full time of overtaking train B to T'Bt. I can't consider it, but
The higher-level control knowledge directly or lower-level control knowledge considers the application of knowledge that deals with changes in vehicle allocation, and selects applicable knowledge that spans between commands and sets application conditions.

(7-c) Strategy knowledge 1560, 1565 is knowledge for controlling upper control buttons R1550, 1555. Which of the upper control buttons tIi1550.1555 to use?
Also, decide how to set parameters and conditions for use. Furthermore, it is also knowledge that changes the application conditions of the higher-level control knowledge 1550° 1555 or switches application to other higher-level control buttons Ta 1550 and 1555 with respect to the inference situation and result. Also, find the cause of the problem that cannot be solved.

Corresponding knowledge is created automatically or by querying humans, and knowledge is evaluated and ranked.

It also includes knowledge to be integrated. Specifically, in addition to this,
This includes knowledge such as checking only important checkpoints, such as delays in trains inspected the next day, and checking only those that are OK later in detail.

The inference unit 16 is a translation unit 16 that changes knowledge into an efficient form.
2. It consists of an interpretation/execution unit 161 that interprets and executes knowledge directly or translated knowledge.

The interface unit 17 inputs data from controlled processes and sensors, and outputs data to these. The process interface section 172 displays the inference process, results, problems, or knowledge in the knowledge section 15,
It consists of a man-machine interface section 171 for inputting data for responses and additional corrections.

In this example, the lower control level and lower levels are homogeneously distributed by function, and the upper control 2 strategy level is hierarchically fractional, so the functions are highly expandable, maintainable, and reliable.
With regard to control, visibility is good and it is possible to control complex and large-scale systems.

This embodiment performs homogeneous decentralization including the control level for relatively low-level control systems or systems that do not require strong organization. For systems with the opposite characteristics, it is easier to perform hierarchical decentralization down to one functional level, and various system requirements can be flexibly dealt with.

FIG. 3 shows a processing flow of one embodiment of the interpretation/execution unit 161. First of all, strategic knowledge! Check 1560-1565 to see if any are executable. This knowledge consists of conditions and conclusions if"the"
Use the n-type heuristic.

At this time, knowledge that satisfies the condition becomes executable knowledge (block 1610). If executable knowledge exists, it is extracted and executed. At this time, if multiple knowledge is executable,
Knowledge to be executed may be selected according to a predetermined priority order (knowledge, knowledge in ascending order of number assigned to knowledge, descending order of location in memory, etc.) (block 1615).
), if there is no executable knowledge, the upper control knowledge 1550
Check ~1555 and see if there is anything executable.
Check as in block 1610 (block 16
11). If there is executable knowledge, the process of block 1615 is performed. If not, lower control knowledge 1540-15
Check 45 and see if there is anything executable (
block 1612).

If there is executable knowledge, perform the process of block 1615; if not, sequentially check the area knowledge below Passenger Command Expertise 1515 to see if there is executable knowledge (block 1613); If so, the process of block 1615 is performed. If there is no executable knowledge, the inference execution ends. After extracting and executing the knowledge in block 1615, the process returns to block 1610 and repeats the above process. When each block of blocks 1610 to 1613 and executable knowledge are found, an execution mode corresponding to the found knowledge is set and executed in block 1615. Block 161 by controlling the number of lX1
It is also possible to improve interpretation and execution efficiency by reducing the overhead of checking 0-1613. Specific examples of controlling the number of execution knowledge include a method of providing a counter in the execution mode, a method of using a timer interrupt, and the like. Also, after executing the executable knowledge at block 1615, the process does not necessarily return to block 1610.

Depending on the execution mode, it is possible to perform flexible and efficient knowledge execution by directly returning to blocks 1611 to 1613, a specific expertise group within block 1613 (for example, train command expertise 1535), or the own block 1615. (block 1616).

〔Effect of the invention〕

As explained above, according to the present invention, specialized knowledge can be divided according to function/control level and distributed hierarchically or homogeneously, and furthermore, it is possible to support the evaluation and modification of other knowledge with knowledge, so it is possible to・In the management system, inference efficiency improves and real-time decision support becomes possible.

2. Reliability ° Since robustness is improved, it becomes possible to support decisions that have a serious impact on human life.

3. It is easy to add and modify knowledge, and its application is flexible, so it is highly practical even when the knowledge itself is unclear and the application situation is often unpredictable.

4. By adjusting the degree of hierarchical distribution and homogeneous distribution, it is possible to flexibly respond to various system requirements.

There are effects such as

[Brief explanation of drawings]

FIG. 1 is an overall block diagram of an embodiment of the present invention, FIG. 2 is a functional block diagram of a system 1 of the present invention, and FIG. 3 is a flow diagram of the main parts of the present invention. l: computer system, 11: firmware storage device, 12: main storage device, 13; auxiliary storage device, 15:
Knowledge Division, 16: Reasoning Division. 171: Man-machine interface department, 1560: Strategy knowledge 1.1550 = Upper control knowledge 1.1540: Lower control knowledge 1.1515: Passenger command expertise, 1525: Train operation command expertise, 1535: Train command expertise, 1
510: Passenger request, 1520: Train schedule knowledge, 153
0: Vehicle schedule knowledge. Agent: Patent Attorney Masaru Ogawa Ni-゛、lZffi

Claims (1)

[Claims] 1. First knowledge for issuing commands in the command system, second knowledge for evaluating, creating, and changing the knowledge, third knowledge for controlling the use of the knowledge, and furthermore, the third knowledge for controlling the use of the knowledge. Create knowledge of 3.
Fourth knowledge to be changed and controlled, dividing the first to fourth knowledge according to each target, function, organization, level, or region of the above directive, and combining these divisions as necessary. A command support method characterized by: 2. Having a knowledge unit that stores divided knowledge, and a reasoning unit independent of the knowledge unit that interprets and executes the knowledge directly or by converting it into efficient knowledge, and is independent of the application field and directives. The instruction/support method according to item 1, characterized in that: 3. The command support method according to item 1, wherein the knowledge is divided hierarchically or homogeneously/equally, and furthermore, both methods are combined as necessary. 4. The instruction/support method described in item 1, which allows for addition, modification, display, and error checking of the above-mentioned knowledge, or for that purpose, which also allows multi-windows. 5. Means for storing first knowledge for issuing commands in the command system, second knowledge for evaluating, creating and changing the knowledge, and third knowledge for controlling the use of the knowledge; means for storing fourth knowledge for creating, changing, and controlling the use of knowledge; means for dividing the first to fourth knowledge according to each target, function, organization, level, or region of the above-mentioned directive; A command support method characterized by having means for combining these divisions as necessary.