JPH0738240B2 - Image-based knowledge-based fault detection system - Google Patents

Description

Detailed Description of the Invention

[Industrial applications]

The present invention relates to an image type knowledge base failure detection device for non-specialists, which performs failure detection processing of plant systems and education and training of operators for failure detection.

[Prior art]

FIG. 3 is a block diagram showing a conventional knowledge-based fault detection apparatus using video as disclosed in Japanese Patent Laid-Open No. 63-761. In FIG. 3, 2 is a rule of an expert for fault detection. A computer equipped with a computer, 5 a television screen on which images, characters, and drawings are displayed, 6 a video disk device equipped with a video of equipment associated with a procedure for finding a failure, and 8 a knowledge-based failure A touch panel for interacting with the discovery device, and 9 is a superimposer for displaying characters on the television screen 5. Next, the operation will be described. First, when a failure occurs in the plant system or the like, the operator selects an item of the failure content from the TV screen 5 by touching an appropriate place on the touch panel 8 with a finger. At this time, when the computer 2 receives the input signal from the touch panel 8, it gives an output command to the superimposer 9 for displaying characters and the video disk device 6 in accordance with a program of rules for failure diagnosis stored in advance. Then, the superimposer 9 gives an output signal for superimposing the two screens of the image of the video disk device 6 and the failure detection operation instruction text to the television screen 5 for simultaneous display. The operator discovers and restores the fault of the plant system based on the instruction statement of the fault finding operation, but on the way, an unknown item (for example, a place to check,
If you have a way to look up, meaning of words, etc.), touch panel 8
By touching the corresponding place above with a finger, a video of the answer (guidance) is displayed on the television screen 5.
In addition, video images that should be the answer (for example, the entire equipment, board appearance, board, relay, chassis card, parts, etc.
All images required for the course in which a specialist discovers a failure) are stored in advance in the video disk device 6.

[Problems to be Solved by the Invention]

Since the conventional knowledge-based failure detection device using video is configured as described above, in order to input the expert's knowledge into this knowledge-based failure detection device, the contents of that knowledge are input as a program as a knowledge base. However, it takes a lot of time. Moreover, there was a problem that it was not possible to easily modify or add rules for inferring a knowledge base. The present invention has been made in order to solve the above-mentioned problems, and in constructing a knowledge base, an operator inputs an enormous amount of knowledge of an expert into a failure detection device without the knowledge of a program. It is an object of the present invention to provide a failure detection device based on an image-based knowledge base that can be easily modified and rules can be easily modified and added.

[Means for Solving the Problems]

An image-based knowledge-based fault detection apparatus according to the present invention is an image input apparatus for reading, as image data, images of a plurality of nodes in which information necessary for inference for fault detection is described. A computer for accumulating the image data read in, and storing the coordinate position on the display screen of each of the nodes,
The display means for displaying an image corresponding to the image data stored in the computer in correspondence with the coordinate position on the display screen, and the connection relation between the nodes in the image displayed on the display screen are displayed on the node display screen. By specifying and deciding the coordinate position of the node, the computer stores the connection relationship between the coordinate positions of the nodes and displays it as an image of the tree-like knowledge structure on the display screen. The computer designated by the coordinate designation device is provided with a coordinate designation device that designates a node used for inference in the image with connection status displayed on the screen by designating the position of the node on the display screen. The node is displayed on the display screen so that it can be distinguished from the non-designated node. In addition, it is equipped with a video disk that stores the video information of the fault-discovered equipment that is associated with the fault-discovery procedure. It is intended to solve unclear points.

[Action]

The expert knowledge is input to the failure detection device according to the present invention by inputting a drawing describing knowledge arranged as a tree-like knowledge structure tree to the computer by the image input device. In addition, by selecting and instructing the relation of the node portion of the tree-like knowledge from the display device via the coordinate designation device, that is, by associating the coordinate position on the image data, the reasoning of the fault discovery is advanced and, if necessary, the television Refer to the screen and operate interactively in sequence to discover a failure.

【Example】

An embodiment of the present invention will be described below with reference to the drawings. In FIG. 1, parts that are the same as those in FIG. 3 are shown with the same reference numerals, and in FIG. 1, 1 is a CRT device (image data (knowledge organized in a tree)) and a CRT device that displays an instruction sentence regarding the operation of the failure detection device ( Display device 3 is a mouse (coordinate designating device) which is a pointing device, and 4 is an image input device which reads and inputs image data such as drawings.
For example, an image scanner or the like. Next, the operation will be described. First, in the video disk device 6, in the same manner as the conventional method, the entire equipment, the board appearance, the board, relays, chassis cards, parts, etc. Video images showing how to conduct surveys by the house are stored. Next, the input of specialized knowledge regarding failure detection will be described. Specialized knowledge about failure detection is organized into a tree, and this is called a knowledge structure tree 7. The knowledge structure tree 7 is traced on the drawing in the form as shown in FIG. In the example of Fig. 2, 7a is the starting point of inference, 7b is a branch (a necessary question on the inference), 7c is an instruction of the investigation, 7d is a conclusion, and the necessary information is shown in text within each frame. Has been done. Since the knowledge structure tree 7 needs to be displayed on the screen of the CRT device 1, if it becomes larger than the screen, the knowledge structure tree 7 is divided into a plurality of sheets, and in FIG. It is represented. The drawing in which the knowledge structure tree 7 is described is directly input as image data through the image input device 4 shown in FIG. 1 and stored in the computer 2. The person who inputs the knowledge is
The image data of the knowledge structure tree 7 is displayed on the CRT device 1, and each node 7a, 7 is displayed according to the message from the system.
Input the connection status between b, 7c, 7d and 7e using the mouse 3. The computer 2 has nodes 7a, 7b, 7c, 7 of the knowledge structure tree 7.
The connection status between d and 7e is stored as the connection status between the coordinate position data designated by the mouse 3. The connection information between the nodes mentioned here indicates a route for finding a failure. Next, the operation for finding the actual failure is performed as follows. That is, if a plant fails, the operator
Search the CRT device 1 for the item of the failure content and select it from the menu using the mouse 3. At this time, the computer 2 receives this input signal and, by the previous input operation, the knowledge structure tree 7 for failure diagnosis accumulated in advance.
Is displayed on the screen of the CRT device 1. At this time, a circle 7f is displayed at the coordinate position of each node recorded by the previous input operation, and the color of the circle indicates the current inference position. The next thing to look for is each node 7a, 7b, 7c, 7
Since the characters are written in the frames d and 7e, the operator inputs a signal using the mouse 3 to the circle 7f in the frame of each node when confirming the contents of the characters. By repeating the above-mentioned operations in sequence, the inference for finding a failure is made. In addition, if there is an unclear point (for example, the place to be examined, the method to be inspected, the meaning of a word, etc.) among the items to be investigated by the operator during the reasoning, the menu on the CRT device 1 should be designated with the mouse 3. If the answer is a video image on the TV screen 5, or if it is a drawing (image data), the CRT
It is displayed on the device 1. In the above example, the circle indicating the position of the node is circled.
Although 7f is used, it may be square or triangular and is not particularly limited. Further, the mark of the frame depending on the meaning of the node is not particularly limited, and any shape may be used. Also. In the above-described embodiment, the procedure for finding a fault in a plant or the like has been described, but the same effect as that in the above-described embodiment can be obtained even if it is used for process diagnosis, control guidance, etc. of various plants.

【The invention's effect】

As described above, according to the present invention, specialized knowledge about plant failure detection and the like is described as a plurality of nodes on the drawing, and the plurality of nodes are input as image data. Since it is recorded as the connection information of the coordinate position data of the node on the image data, a large amount of specialized knowledge data can be input easily and quickly without the knowledge of the program, and the modification or addition of knowledge is also easy. There is an effect that can be done.

[Brief description of drawings]

FIG. 1 is a block diagram showing a knowledge-based fault detection apparatus according to an embodiment of the present invention, FIG. 2 is a model diagram of a knowledge structure tree according to the present invention, and FIG. 3 is a conventional knowledge-based fault detection apparatus. It is a block diagram shown. 1 is a CRT device (display device), 2 is a computer, 3 is a mouse (coordinate designating device), 4 is an image input device, 5 is a television screen, 6 is a video disk device, and 7 is a knowledge structure tree. In the drawings, the same reference numerals indicate the same or corresponding parts.

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Takayuki Tanaka 1-2-2 Wadasaki-cho, Hyogo-ku, Kobe-shi, Hyogo Mitsubishi Electric Corporation Control Works (72) Inventor Shingo Aya, Hyogo-ku, Kobe-shi, Hyogo 1-2-1 Tasakicho Mitsubishi Electric Co., Ltd. Control Works (72) Inventor Megumi Nonoyama 1-2-2 Wadazakicho, Hyogo-ku, Kobe Hyogo Prefecture Mitsubishi Electric Co., Ltd. Control Works (56) References Special Kai 63-20529 (JP, A) JP 63-291130 (JP, A)

Claims (2)

[Claims] 1. An image input device for reading, as image data, images of a plurality of nodes in which information necessary for inference for finding a failure is written, and image data read by the image input device is stored. A computer for storing the coordinate position on the display screen of each of the nodes; display means for displaying an image corresponding to image data stored in the computer in association with the coordinate position on the display screen; By determining the connection relationship between the nodes in the image displayed on the screen by designating the coordinate position of the node on the display screen, the computer stores the connection relationship between the coordinate positions of the nodes. Displayed on the display screen as an image of a tree-like knowledge structure, and
A coordinate designating device for designating a node used for inference in the image with connection status displayed on the display screen for fault detection by designating a position of the node on the display screen; The fault finding device based on an image-based knowledge base, wherein the computer displays a node designated by the coordinate designation device on the display screen so that the node can be distinguished from a non-designated node. 2. A video disc for storing an image of a fault-discovered equipment associated with a fault-discovery procedure, and information stored in the video disc when there is an unclear point for the fault-discovery. The fault finding device based on the image-type knowledge base according to claim 1, wherein the unknown points are solved by using the.