JP2828973B2 - Power system accident judgment device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Purpose of the Invention] (Industrial application field) The present invention automatically determines an accident section, aspect and content of an electric power system based on information observed from an electric power system. The present invention relates to a power system accident judging device for making a judgment. (Prior Art) At present, judgment of an accident in a power system largely depends on human judgment. In recent years, a technique for inferring an accident section, an accident aspect, an improper operation of a protective relay or a circuit breaker, and the like has been developed. This kind of technique is usually performed using accident determination knowledge. And the knowledge of accident determination in the power system includes
Basic knowledge, such as the operating principle of protection relays, and empirical aspects such as those that exceed the ability of human judgment prediction in a short time, those that involve wide-range interruptions, and those that involve unauthorized operation of equipment. There is strong knowledge. Many of the so-called serious accidents belong to the latter knowledge. Conventionally, when an accident is determined using such knowledge, a predetermined knowledge is sequentially incorporated into a previously formed determination logic. (Problems to be Solved by the Invention) As described above, when an accident is determined by the conventional device, the logic of the determination and the determination knowledge are fixed, and the combination of these is also fixed. Is missing. Further, there is a problem that it is impossible to add the empirical knowledge and to correct the already input knowledge, so that it is impossible to increase the accident determination ability. The present invention has been made to solve the above problems, and when an accident occurs in the power system, not only a simple accident,
It is an object of the present invention to provide a power system fault judging device provided with a flexible fault judging method for occurrence of multiple faults, malfunctions, and complex faults accompanied by malfunctions. [Structure of the Invention] (Means for Solving the Problems) The present invention will be described with reference to FIG. 1. Control knowledge 211, decision knowledge 212 consisting of empirical knowledge about the operating principle of a protection relay used in a power system and a system fault, and basic matters 213 consisting of a plurality of elements for individually specifying the contents of this decision knowledge. And a plurality of inference methods by the inference engine 22 are provided. (Operation) An accident determination is made based on a determination order determined by the control knowledge 211, a knowledge group used for the determination, and an inference method. Here, if each knowledge is individually edited using the editing functions 231 to 233, the combination of these respective knowledges becomes extremely large,
As a result, it is possible to perform a knowledge-based flexible accident determination. (Example) Hereinafter, an example is described with reference to drawings. FIG. 1 is a functional block diagram of an embodiment of a power system accident determination apparatus according to the present invention. In FIG. 1, 1 is systematic change information, 2 is a power system accident determination device, and 3 is a determination result. The power system accident determination device 2 includes a knowledge base 21 including control knowledge 211, determination knowledge 212, and basic matters 213, an inference engine 22 including a plurality of inference methods (inference method 1, inference method 2,... Inference method n),
It is composed of a plurality of editing functions 231 to 233. Here, as shown in FIG. 2 (a), the control knowledge consists of an application order for each adopted knowledge name, an inference method, etc., and the judgment knowledge is a module as shown in FIG. 2 (b). Knowledge groups (judgment knowledge module 1, judgment knowledge module 2,...)
It consists of a judgment knowledge module m), the details of which will be described later. FIG. 3 is a detailed configuration diagram of the knowledge base. An example of this knowledge is shown below. Control knowledge (operation principle, 1, in2, [operation relay],
[Simple accident]), in the above knowledge description, “operation principle” is the name of the judgment knowledge to be adopted for judgment, “1” is the order of application of the judgment knowledge module “operation principle” during inference,
“In2” is an inference method (described later) in the “operation principle” module, “[operation relay]” is an input value when applying the “operation principle” knowledge, and “[simple accident]” is an output value. This means that "the decision knowledge module called" operation principle "is adopted first in the system fault point inference, and the inference method at that time uses" in2 "(search in order of priority and apply one). The input to the "principle" is an "operation relay" and the output is a "simple accident.", And is described in the PROLOG language. Next, there are four judgment knowledge modules for judgment knowledge. "Operating principle" means the operation principle of the relay and the knowledge to extract the accident candidate section. "Multiple accidents" means the knowledge to extract the possibility of multiple accidents occurring simultaneously from the accident candidate section.
The “malfunction / non-operation accident” is knowledge for determining an accident involving a malfunction of a relay or a circuit breaker, and the “malfunction knowledge” is knowledge for determining an accident involving a malfunction of a device. Note that simple accidents, multiple accidents, malfunctions and malfunctions, etc. are referred to as accident types. Among them, knowledge of the operation principle will be described. Operation principle (-operation relay, simple accident): -P0 (-priority, 1), P1 (-operation relay, -interruption CB), P2 (-interruption CB, -interval), P3 (-interval, -simple accident) ). In the above knowledge description, “−operation relay” is an input value,
“−simple accident” is an output value. “P0” to “P3” are basic knowledge defined in the basic matter 213, “P0 (−priority, 1)” indicates that the priority of this knowledge is 1, and “P1 (− "Operation relay, -Interruption CB)" indicates that the CB interrupted by the activated relay is -Interruption CB, and "P2 (-Interruption CB, -Section)" indicates that the load-side equipment of -Interruption CB is -Section. And "P3 (-section, -simple accident)" indicates that the power failure in -section is -simple accident. This means that, of the CB load-side facilities cut off by a certain operation relay, the facility that has lost power is a simple accident section (priority = 1).
It is described in OLOG language. The individual basic knowledge “P0 (−priority,
1), "P1 (-operation relay, cut-off CB)", "P2 (-
The combination of "interruption CB, -section)" and "P3 (-section, simple accident)" is called a knowledge set unit for convenience. Fig. 4 shows an example of the configuration of an inference engine. In this example, there are three inference methods. Here, “all search” is a method of searching for knowledge modules and extracting all applicable ones, and “priority order” is priority order (smaller numbers indicate higher priority). Is a method in which inference is completed when one applicable application is extracted by exploring the knowledge module. In addition, the “ordering on the knowledge base” is a method in which knowledge sets are arranged in the order in which the knowledge modules are arranged. This is a method in which inference is terminated when a unit is searched and one applicable unit is extracted, as shown in the control knowledge of FIG. Motion accident → multiple accidents →
Inference is performed using the inference engine in the order of malfunctions, and a determination result is output. The above-described series of processing contents will be described in detail. First, the operation principle knowledge (operation principle 1) is searched in order of priority (in2) by the operation relay [operation relay] to extract a simple accident section [simple accident], and the malfunction / non-operation knowledge (misoperation error 2) (In1) to extract all the possible malfunctions and malfunctions in the simple accident section [simple accident, multiple accidents]. The accident knowledge (multiple accidents 3) is searched in order of priority (in2), and finally, the simple
For multiple accidents, the malfunction knowledge is searched in the order of the knowledge base (malfunction accident 4) (in3), and the possibility of malfunction is extracted and output as a judgment result. In this inference process, the basic knowledge constituting the knowledge set unit is checked individually, and an accident determination is performed according to each knowledge set unit. Next, the editing function will be described. The editing function 231 edits the control knowledge 211, and can add / delete the control knowledge, modify the content of the control knowledge (judgment knowledge name, application order, inference method, input value, output value), and search. An example in which the application order and the inference method of the judgment knowledge module described in the control knowledge of FIG. 3 are changed is shown below. As a result, the order of application of the knowledge module for judging multiple accidents at fault inference and false malfunction accidents is changed (multiple accident 3
2. Malfunction failure accidents 2 → 3) Investigation method of knowledge of malfunction failure accidents should be searched in order of priority and one should be applied (in1 → in)
2) Therefore, if the control function is edited as follows, the judgment order is as follows: operation principle → multiple accidents → malfunctioning accidents → malfunctioning accidents. You will find one applied knowledge. The editing function 232 edits the judgment knowledge 212 and can add / delete the judgment knowledge, modify the contents of the judgment knowledge module (combination of basic items and variables of the basic items), and search. An example in which the judgment knowledge of FIG. 3 is modified is shown below. The above "P4 (-Shutdown CB, -Section)" means -Shutdown
This indicates that the CB power supply equipment is in the-section, and that `` Of the CB power supply equipment cut off by a certain operation relay,
The facility where the power failure has occurred is the simple accident section. (Priority level 2). " Therefore, if the judgment knowledge is edited as follows, the inference method of the operating principle knowledge is a method of exploring in order of priority. Therefore, by changing the priority from 1 to 2, even if this knowledge is applied before editing, If, after editing, there is something that can be applied with higher priority knowledge, this knowledge will no longer be applied. Also, the basics of knowledge content are P3
To P4, and even if this knowledge is searched and applied, the content of the knowledge application result (-simple accident) is different from that before editing. Editing function 233 edits basic items and adds basic knowledge.
It is possible to delete, modify the contents of basic knowledge (article definition in PROLOG language), and search. An example of correction of the basic knowledge in FIG. 3 is shown below. P0 (X, Y): -X ¥ == Y .... (￥ == is an operator indicating that they are not equal). Thus, "P0 (X, Y)" is changed from the first variable: X to the second variable Y. Becomes the basic knowledge that means examining inequality. Therefore, if you edit the basic knowledge as
(-Priority 1) "means" priority is not 1 ". In the case of knowledge to be subjected to the inference method of the priority order, if it is before editing," P0 (-priority 1) " The knowledge that keeps "degree, 1)" is searched earlier than the knowledge of "P0 (-priority, 2)", but after editing, the search order is reversed. As described above, when the knowledge base is individually edited using the editing functions 231 to 233, various determination methods can be performed, and the determination efficiency and the determination result are affected. As another embodiment, the determination knowledge 211 of the knowledge base 21 may be set as follows. Control knowledge (judgment knowledge name, order of application, inference method, limit priority, input, output), here new "limit priority"
Add. This is an item that is effective in a method of searching in order of priority and inferring until one piece of knowledge can be applied (referred to as in2). When inferring by in2, only knowledge having a priority equal to or higher than this limit priority is searched. set to target. Thereby, when inference is performed by in2 for the judgment knowledge module having the default knowledge with the lowest priority in the judgment knowledge module, if this limit priority is lower than the priority of the default knowledge, the inference is performed at the default knowledge. It always succeeds. Conversely, if it is set high, the inference may fail. By adding the marginal priority in this way, the success / failure of the inference itself can be controlled. Further, as another embodiment, the judgment knowledge 212 of the knowledge base 21 is used.
It is also conceivable to add a judgment knowledge module for "past accident experience" to the above. Here is “past accident experience”
Is past empirical knowledge about an accident that is difficult to determine from the operating characteristics of the relay alone. By applying this module after the operation principle, an accident section can be more reliably extracted. [Effects of the Invention] As described above, according to the present invention, the rules of the knowledge base have a three-layer structure, and the execution order of the knowledge group of the judgment knowledge and the inference method can be specified. In addition to the configuration based on items, a plurality of inference methods are provided at the time of inference, and the combination thereof is designated by control knowledge, so that the following effects can be obtained. Inference processing can be easily controlled, and an inference method suited to the problem can be configured. You can divide the problem into sub-problems,
A suitable inference method can be selected for each subproblem. It becomes possible to systematize the judgment knowledge, and it is possible to easily make corrections and changes. It is easy to maintain the consistency of the entire knowledge base (such as consistency between rules). As a result, it is possible to construct a practical knowledge-based system in the field of "power system accident determination", and to perform the analysis and analysis of the cause of the accident much faster than experts. It became so.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a functional block diagram of an embodiment of a power system accident judging device according to the present invention, FIG. 2 is a schematic diagram of control knowledge and judgment knowledge, and FIG. 3 is a detailed configuration of a knowledge base. FIG. 4 shows the configuration of the inference engine. 1. System information change information 2. Power system accident determination device, 21
…… Knowledge base, 22 …… Inference engine, 231-233 …… Editing function.

Claims (1)

(57) [Claims] Priority of inference applied to each of a plurality of knowledge modules based on empirical knowledge of protection relays used in power systems and multiple accidents, malfunctions, malfunctions, etc. A judgment knowledge having a plurality of knowledge set units configured by combining basic knowledge such as a circuit breaker and an accident section corresponding to the circuit breaker, and an inference order, an inference method, an input value, and an output value are described for each judgment knowledge module. A knowledge base composed of a plurality of control knowledges, and basic matters described by the basic knowledge, and an inference method described in the control knowledge, by searching for a knowledge set unit constituting a determination knowledge module and applying the search. An inference method that stops all the knowledge set units and stops the inference. If the knowledge set units are searched in order of priority and applicable ones are extracted An inference engine having an inference method for stopping the inference and an inference method for stopping the inference if an applicable one is extracted by searching for a knowledge set unit in the order of arrangement on the knowledge base. A first editing unit that can independently edit the inference order or input values, output values, or applied inference methods of the determination knowledge module; provided in correspondence with the determination knowledge; A second editing unit that can independently edit the knowledge module; and a third editing unit that is provided corresponding to the basic matter and that can independently edit the basic knowledge, and is specified by the control knowledge. A power system accident judging device characterized by performing inference according to the determined application sequence of each judgment knowledge module and outputting a judgment result.