JPH06149826A - Support device for plant management - Google Patents

Abstract

PURPOSE:To generally consider the decision of evaluation of plants with no individual difference and to decide these plants on the same evaluation standard by generating automatically a fuzzy production rule and a membership function and carrying out a fuzzy operation to decide the evaluation marks of each plant. CONSTITUTION:A fuzzy operation is carried out based on the final working data obtained by an input device 1, a fuzzy production rule, and a membership function. That is, the narks calculated by a CPU 3 are totalized and displayed on a screen for each plant. For instance, the facilities, the quality and the production quantity have each reference point and consist of two or more pieces of information having the proper reference points in response to each importance as the output examples of a CRT display device 4. Then the evaluation marks of the alarm information, the facilities information, and the working information are calculated by the fuzzy operation. The device 4 can display the additional data necessary for the management items in addition to the evaluation marks and also has a data selection function to control the importance to the functions.

Description

Detailed Description of the Invention

[0001]

The present invention is used as a computer-aided device for managing and operating a production factory. The present invention relates to an apparatus that assists in organizing target data for factory operation and evaluating an operation state.

[0002]

2. Description of the Related Art Although a production factory is managed and operated by a person (manager, manager), various information generated from the factory is used in order for the "person" to know the actual operating state of the factory. To be This information is alarm information, equipment information, operation information, quality information, complaint information, production information, inventory information, sales information, and the like. These pieces of information have complexly related contents rather than independent phenomena. In addition, these kinds of information are extremely diverse in a real factory. For example, even if attention is paid to alarm information, there are many kinds of alarm information generated from one production line, and if an alarm generated once is uniformly treated and evaluated for all alarms, accurate information evaluation cannot be performed.

In order to make this easy to understand, it is widely practiced to process or accumulate information for each management item so that each person in charge can easily judge it and output the information as a graph or trend. ing. Supporting devices in that sense have been known. These organized results will be submitted to the meeting for management evaluation. Regarding the evaluation of the management submitted and considered at the meeting, the judgment criteria depend on people.

[0004]

Since each person has different experience and thought, it is unavoidable that there are individual differences in judgment criteria.

When a large-scale company has a plurality of similar production plants all over the country and various types of information obtained from the factories of various places are organized to evaluate the operating state of each plant, the same person is the same. It is impossible to evaluate the whole with the evaluation standard of. This means that the same evaluation criteria cannot be applied when there are multiple factories. Even if the evaluation criteria are set and shown on the paper, the content is complicated and cannot be constant.

[0006] Further, there are cases in which a person cannot understand even the subtle causal relationship of the information obtained, and it may not be possible to make an appropriate judgment. In other words, it is possible for an evaluation to be performed without knowing which item is the true main cause.

The present invention solves these problems, and the evaluation judgment of information in factory operation can be comprehensively considered without individual difference, and a support device for judging a plurality of factories by the same evaluation standard is provided. The purpose is to provide.

[0008]

The present invention provides a central processing unit, a database connected to the central processing unit, an output device such as a CRT display device connected to the central processing unit, and the central processing unit. It has a connected input device, and the fuzzy production rules and the membership functions corresponding to it are accumulated in the database, and the information collected and processed by the input device and the corresponding fuzzy production rules and membership functions are automatically created. Then, a fuzzy operation is performed on them, and the centroid point of the operation result is calculated to determine the evaluation point. The calculation standard for this evaluation score can be automatically adjusted according to the factory condition and / or factory management goals. Also,
The evaluation points and the information attached thereto are displayed on the CRT display device, and the operation status of each factory can be grasped at a glance.

Further, the evaluation points are weighted and averaged for each facility (alarm information, facility information, operation information), quality (quality information, complaint information), production progress (production information, inventory information, sales information), Scores can be used as three criteria in factory operations.

Further, each weight of the weighted average has a function of being able to be sequentially adjusted according to the state of the factory at that time and / or the factory management target, and one of the feelings about the points displayed on the CRT display device can be selected. By selecting, the ideal weight at that time is automatically determined. The weight determination method uses the selected sensation as input data, calculates the ideal evaluation point by means having the same function as the fuzzy used when determining the evaluation point, and calculates it by the back propagation method in the neural network. It is characterized by having a function of adopting the result of learning so as to approach the ideal evaluation point.

Further, specifically, the information about the equipment is
The cumulative number of occurrences and / or cumulative duration weighted according to the importance of each type of alarm information generated from the equipment during the set period can be used. It is possible to use the cumulative number of detections and / or the cumulative duration weighted according to the importance of each target device of the operation level information detected by the facility. It is also possible to use the number of times of abnormality detection. Further, the information on the facility can be an average utilization rate obtained by weighting the utilization rate calculated from the time the facility has been operating during the set period according to the importance of the facility.

[0012] The information regarding the quality can be the cumulative number, which is weighted according to the degree of the out-of-specification number of the product that has occurred during the set period. Also,
The quality information can be the cumulative number of complaints weighted for each product regarding the complaints from the consumer of the product generated during the set period.

The information regarding the production quantity can be an average achievement rate obtained by weighting the production achievement rate with respect to the planned production quantity during the set period, and the information regarding the production quantity can be obtained during the set period. The out-of-stock quantity and the out-of-stock quantity exceeding the inventory overflow threshold can be set as cumulative quantities weighted by product type. The information regarding the production quantity may be an average achievement rate obtained by weighting the sales achievement achievement rate with respect to the planned sales quantity during the set period by the product type.

A fuzzy set is an extended concept of a set introduced to express an ambiguous concept, and a universal set U
The mapping from to the interval [0, 1] is specified by the membership function μ. The significance of the membership function is described in Iwanami Shoten "Information Science Dictionary" and others.

Backpropagation is one of the typical methods of neural networks, which is an attempt to imitate human neural circuits as information processing, so that an ideal output value called a teacher signal can be calculated. It is a method of adjusting the structure coefficient that constitutes the network model. Backpropagation (error backpropagation) is described in detail in "Neural Network Information Processing", Sangyo Tosho Co., Ltd. (June 20, 1988), pp. 50-54.

[0016]

[Function] The fuzzy production rules corresponding to the management items of the factory operation are registered in Japanese in the database, and the membership function corresponding to each rule is numerically registered. Further, information for each management item such as alarm information, equipment information, operation information, quality information, complaint information, production information, inventory information, and sales information that are totalized and processed by the input device is stored in another database. The central processing unit retrieves the information for each management item, automatically creates a fuzzy production rule and a membership function corresponding to the information, and performs a fuzzy operation according to the rule. The calculation result is represented by a figure, its center of gravity is calculated, and the value of its x coordinate is converted into a score of 100 points. This score is stored in a database that stores information for each management item. Further, when the importance of each factor of the management item becomes inappropriate in the factory operation, it is automatically set to an ideal value by learning by the neural network. This allows users to use C when they want to use it.
The information necessary for factory operation can be checked and examined on the RT display device, and the evaluation of factory operation can be automatically adjusted according to the purpose of the user.

[0017]

Embodiments of the present invention will now be described with reference to the drawings. FIG. 1 is a block diagram showing the basic configuration of the embodiment of the present invention. The present invention comprises an input device 1, a database 2 in which fuzzy production rules and membership functions are registered, a central processing unit 3 for performing fuzzy operations, and a CRT display device 4 connected to the central processing unit 3.

The operation of each means of the apparatus of the embodiment of the present invention configured as above will be described.

In the database 2, alarm information, equipment information, operation information, quality information, complaint information, production information, inventory information, sales information are recorded, and each fuzzy production rule and membership function are registered. The central processing unit 3 searches the database 2 for fuzzy production rules and membership functions corresponding to the information (management items) tabulated and processed by the input device 1 and automatically creates fuzzy calculationable information. For fuzzy operations and membership functions, see "Fuzzy Computer-Fuzzy Expert System" published by Trikeps Co., Ltd.
(July 28, 1988) Pages 26 to 30 Published by Management Development Center, "Fuzzy System Practical Theory and Applied Cases" (January 8, 2001) No. 10
See page 30 to page 30 for details.

Further, the central processing unit 3 uses the information obtained from the input device 1 and the information obtained from the database 2 to perform a fuzzy operation. Next, the central processing unit 3 represents the result obtained by the fuzzy calculation by a figure, and calculates a value obtained by normalizing the x coordinate of the center of gravity of the figure from 0 to 100 as an evaluation point. Also, the CRT display device 4
The evaluation points calculated for each management item by the central processing unit 3 are weighted and averaged as evaluation points for each of equipment, quality and production progress, and displayed as three judgment criteria in factory operation. This weighting is adjusted according to the state of the factory at that time and / or factory management goals.

Next, a detailed processing operation of the embodiment of the present invention will be described with respect to alarm information in equipment.

First, the processing operation performed by the input device 1 will be described with reference to data. FIG. 2 is a flow chart showing the flow of processing operations of the input device 1 in the embodiment of the present invention. First, information about alarms that occurred during the day (number of interruptions, interruption time, number of alarms) is totaled. At the same time, an alarm is generated in the plants A, B, and C, and the details of the alarm are as follows.・ Plant A has 30 interruptions and 200 interruptions
Minutes, 100 alarms. -Plant B has 20 interruptions and 100 interruptions
Minutes, 200 alarms.・ Plant C was interrupted 10 times, interrupted for 80 minutes,
The number of warnings is 150 times. Here, assuming that the importance of plants A, B, and C for the entire plant is 0.8, 0.6, and 0.3, the primary processing data are as follows.

[0023]

[Equation 1] Here, the number of interruptions in the area evaluated by the primary machining data is up to 100 times, the interruption time is up to 600 minutes, and the number of alarms is 1
The secondary processing data is as follows when the number of times is up to 000.

Number of secondary machining interruptions = X ₁ ÷ 100 Number of secondary machining interruptions = X ₂ ÷ 600 Number of secondary machining alarms = X ₃ ÷ 1000 Further, in order to eliminate the disparity in scale between factories, it is called the number of plant weighting plants. The value obtained by dividing the secondary machining data by the value obtained by summing the importance of the plants for each factory is the final machining data of the input device.

Next, the central processing unit 3 will be described. FIG. 3 is a flow chart showing the flow of processing operations of the central processing unit 3 in the embodiment of the present invention.

First, a fuzzy production rule corresponding to the final processed data of the input device 1 is created as follows. 1. If the number of interruptions is more than a few and the interruption time is long, the equipment is very bad. 2. If the number of interruptions is large, the interruption time is short and the number of alarms is large, the equipment is very bad. 3. If the number of interruptions is large, the interruption time is short and the number of alarms is very small, the equipment is bad. 4. If there are a few interruptions, a short interruption time and very few alarms, the equipment is a little bad. 5. If the number of interruptions is small, the interruption time is short and the number of alarms is large, the equipment is bad. 6. If the number of interruptions is small, the interruption time is short and the number of alarms is very large, the equipment is very bad. 7. If there are no interruptions and very few alarms, the equipment is the best. 8. If there are no interruptions and many alarms, the equipment is good. 9. If there are no interruptions and the number of alarms is very high, the equipment is a little bad.

Further, the membership function corresponding to this rule is automatically assigned. The membership function groups are shown in FIGS. 4 and 5. Here, the fuzzy operation is performed by the final processed data by the input device 1, the fuzzy production rule, and the membership function. However, for simplicity of explanation, it is assumed that there are the following two rules. 1. When the number of interruptions is high, the interruption time is short, and the number of alarms is low, the equipment is bad. 2. When the number of interruptions is small, the interruption time is short, and the number of alarms is very small, the equipment is a little bad. The fuzzy operation for the above two rules is as shown in FIG. Here, since the x-coordinate of the center of gravity is calculated with a value between -1.0 and 1.0, a value normalized from 0 to 100 is used as the evaluation point.

Next, the CRT display device 4 will be described. First,
The points calculated by the central processing unit 3 are totaled and displayed on the screen for each factory. FIG. 7 shows a CR in the embodiment of the present invention.
It is an output example of the T display device 4. Each of the equipment, quality, and production quantity has 100 points as reference points, and these are composed of two or more pieces of information having unique reference points according to their importance. For example, in the evaluation of equipment, there are 20 unique management points for each of the three current management items, that is, alarm information, equipment information, and operation information.
The points are 30, 30, and 50. Here, the evaluation points of the alarm information, the facility information, and the operation information are calculated as 80 points, 40 points, and 84 points, respectively, by the above fuzzy calculation,
Therefore, the equipment has 16 points (20 x 0.8
), 12 points (30 × 0.4), 42 points (50 × 0.84), and the total score for equipment is 70 points. Also this C
The RT display device 4 can display not only evaluation points but also attached data necessary for management items, and data selection for adjusting the importance of each factor (equipment, quality, production quantity) of management items. have. We support decision making in factory operations from these display data.

FIG. 8 is a flow chart showing the arithmetic logic for the information relating to the production quantity according to claims 7 and 9. In other words, it is information calculated from the actual production quantity against the planned production quantity during the set period, and is the logic to calculate the cumulative number of unachieved products and / or the average achievement rate of unachieved products weighted by product type or set. This is information calculated from the actual sales quantity against the planned sales quantity during the period, and shows the logic for obtaining the average number of unachieved products and / or the cumulative achievement rate of unachieved products weighted by product type. FIG. 8 is common to the two logics.

FIG. 9 shows calculation criteria (importance for each factor)
Is a necessary selection item for adjusting. That is, when it is determined that the decision making in the factory operation cannot be made smoothly according to the currently set calculation criteria, the optimal feeling for the score state is selected from the nine items in FIG. It is adjusted to the ideal calculation standard.

FIG. 10 shows a neural network model for performing the above adjusting function. In FIG. 10, a circle indicates a unit, and a circle indicates that the factors in the input layer and the intermediate layer are all related to each other. The input value is the evaluation score of each item, and is input in a fixed sequence in the order of equipment, quality, and production quantity. Then, the coupling coefficient between the units is changed (learned) until the output value approaches the ideal point (teaching signal) calculated from the above feeling, and the coupling coefficient between the middle layer and the output layer at the end of learning is the importance of each item. Becomes As the method of changing the structure coefficient, a value proportional to the difference between the output value and the ideal point and the current coefficient is adopted as the amount of change, as in the general back propagation method. The calculation method in each unit is to apply a logistic function to the sum of the values obtained by multiplying the output value of the previous layer in the unit and the coupling coefficient. For the significance of units and logistic functions, see "Neural Network Information Processing" (1
(June 20, 1988) Item 7 to 14 have detailed description.

Further, an embodiment of the automatic adjustment function of the present invention will be described. For example, the evaluation score of the equipment shown in FIG. 7 is 7
When the score is 0, if the user wants to focus on managing the equipment, it is necessary to lower the evaluation score level. In other words, the lower the evaluation score, the more factory resources (human resources, equipment, information, funds) will be provided to the equipment. Therefore, in this case, if 2 is selected from the 9 senses of FIG. 9,
An ideal point matching the degree of the sense is calculated by fuzzy calculation. In this case, it was calculated as 60 points. The 60 ideal points are the ideal points of the equipment shown in FIG. 10, that is, the teacher signals. Next, the coupling coefficient between the units is changed by the above-described changing method until the total point of the equipment approaches the ideal point of the equipment by the back propagation method in the neural network. Then, the coefficient between the middle layer and the output layer after the change becomes the importance of the new management item.

[0033]

As described above, according to the present invention, the state of production activity of each factory can be represented by the same evaluation axis as support for factory operation, and the establishment of factory management technology can be aimed at. Therefore, in the factory strategy, there is an effect that the user can optimally allocate the resources such as human resources (knowledge), equipment, information, funds, and the like to the factory where the evaluation is low and effectively operate.

[Brief description of drawings]

FIG. 1 is a block diagram showing a basic configuration of an embodiment of the present invention.

FIG. 2 is a flowchart showing the flow of processing operations of the input device according to the embodiment of the present invention.

FIG. 3 is a flowchart showing the flow of processing operations of the central processing unit in the embodiment of the present invention.

FIG. 4 is a membership function in alarm information evaluation according to the embodiment of the present invention.

FIG. 5 is a membership function in alarm information evaluation according to the embodiment of the present invention.

FIG. 6 is an example of fuzzy calculation by virtual data according to the embodiment of the present invention.

FIG. 7 shows an output example of the CRT display device in the embodiment of the present invention.

FIG. 8 is a diagram showing an example of a calculation logic of information regarding a production quantity in the embodiment of the present invention.

FIG. 9 is a feeling required for adjusting management item importance in the embodiment of the present invention.

FIG. 10 shows an example of a neural network model in the embodiment of the present invention.

[Explanation of symbols]

 1 input device 2 database 3 central processing unit 4 CRT display device

─────────────────────────────────────────────────── ───

[Procedure amendment]

[Submission date] December 28, 1992

[Procedure Amendment 1]

[Document name to be amended] Statement

[Name of item to be corrected] 0003

[Correction method] Change

[Correction content]

In order to make this easy to understand, it is widely practiced to process or accumulate information for each management item so that each person in charge can easily judge it and output the information as a graph or trend. ing. Supporting devices in that sense have been known. These organized results, will be subject to evaluation of the manager over impingement been submitted to the conference. The criteria also for the evaluation of the manager over impingement to be submitted to the meeting examined are those that merely rely on the people.

[Procedure Amendment 2]

[Document name to be amended] Statement

[Correction target item name] 0010

[Correction method] Change

[Correction content]

Further, each weight of the weighted average has a function of being able to be sequentially adjusted according to the state of the factory at that time and / or the factory management target, and one of the feelings about the points displayed on the CRT display device can be selected. By selecting, the ideal weight at that time is automatically determined. The method of determining the weights were as input data the selected sensory calculates the ideal evaluation point means having the same function as fuzzy adopting during the evaluation point determined by a back propagation method in the neural network, the It is characterized by having a function of adopting the result of learning so as to approach the calculated ideal evaluation point.

[Procedure 3]

[Document name to be amended] Statement

[Correction target item name] 0012

[Correction method] Change

[Correction content]

[0012] The information regarding the quality can be the cumulative number, which is weighted according to the degree of the out-of-specification number of the product that has occurred during the set period. Also,
Information about the quality, can be attached to complaints from consumers the products in a time that is set, the cumulative number claims weighted for respective product.

[Procedure amendment 4]

[Document name to be amended] Statement

[Name of item to be corrected] 0015

[Correction method] Change

[Correction content]

Backpropagation is one of the typical methods of neural networks, which is an attempt to imitate human neural circuits as information processing, so that an ideal output value called a teacher signal can be calculated. It is a method of adjusting the coupling coefficient that constitutes the network model. Backpropagation (error backpropagation) is described in detail in "Neural Network Information Processing", Sangyo Tosho Co., Ltd. (June 20, 1988), pp. 50-54.

[Procedure Amendment 5]

[Document name to be amended] Statement

[Name of item to be corrected] 0024

[Correction method] Change

[Correction content]

Secondary machining interruption count = X ₁ ÷ 100 Secondary machining interruption time = X ₂ ÷ 600 Secondary machining alarm count = X ₃ ÷ 1000 Further, in order to eliminate the difference in scale between factories, it is called the number of plant weighting plants. The value obtained by dividing the secondary machining data by the value obtained by summing the importance of the plants for each factory is the final machining data of the input device.

[Procedure correction 6]

[Document name to be amended] Statement

[Correction target item name] 0031

[Correction method] Change

[Correction content]

FIG. 10 shows a neural network model for performing the above adjusting function. In FIG. 10, a circle indicates a unit, and a circle indicates that the factors in the input layer and the intermediate layer are all related to each other. The input value is the evaluation score of each item, and is input in a fixed sequence in the order of equipment, quality, and production quantity. Then, the coupling coefficient between the units is changed (learned) until the output value approaches the ideal point (teaching signal) calculated from the above feeling, and the coupling coefficient between the middle layer and the output layer at the end of learning is the importance of each item. Becomes As a method of changing the coupling coefficient, a value proportional to the difference between the output value and the ideal point and the current output value is adopted as the amount of change, as in the general back propagation method. The calculation method in each unit is to apply a logistic function to the sum of the values obtained by multiplying the output value of the previous layer in the unit and the coupling coefficient. For the significance of units and logistic functions, see "Neural Network Information Processing" published by Sangyo Tosho Co.
(June 20, 1988) Sections 7 to 14 have detailed descriptions.

Claims (9)

[Claims] 1. A data input means, a table holding a function, a calculation means for calculating data given to the input means by referring to the table, and an output means for outputting a calculation result of the calculation means. In the supporting device for factory operation provided with, the data given to the input means includes (a) two or more pieces of information about facilities in the factory, (b) two or more pieces of information about product quality of the factory, (C) two or more pieces of information about the production quantity of the factory,
Data that includes one or more of the above, and the function held in the table is a membership function defined by a fuzzy set, and is a function that evaluates the degree of the operating state of the factory from the data. The calculation means is a fuzzy inference calculation means, and the output of the output means is a simple evaluation score, and the calculation standard of the score can be automatically adjusted according to the state of the factory and / or the factory management target. Supporting device for factory operation. 2. The factory according to claim 1, wherein the information on the facility is a cumulative number of occurrences and / or a cumulative duration weighted according to the degree of importance of each type of alarm information generated from the facility during a set period. Supporting device for operation. 3. The factory operation according to claim 1, wherein the information on the facility is a cumulative number of detections weighted according to the importance of each target device of the operation level information detected by the facility during the set period. Support equipment for. 4. The facility-related information is calculated from an average utilization rate obtained by weighting the utilization rate calculated from the time the facility has been operating during a set period according to its importance and / or the production amount and the operating time. 2. The support device for factory operation according to claim 1, wherein the capability is an average capability weighted by facility according to its importance. 5. The support device for factory operation according to claim 1, wherein the information on the quality is a cumulative number of the standard deviations of the product generated during a set period, weighted according to the degree. 6. The support apparatus for factory operation according to claim 1, wherein the quality information is a cumulative number of complaints weighted for each product with respect to complaints from the consumer of the product generated during a set period. . 7. The information on the production quantity is information calculated from the actual production quantity with respect to the planned production quantity in the set period, and is the cumulative quantity of unachieved products and / or unachieved products weighted by product type. The support device for factory operation according to claim 1, which is an average achievement rate. 8. The factory operation according to claim 1, wherein the information on the production quantity is a cumulative quantity weighted by a product type for each of the out-of-stock quantity and the over-stock quantity exceeding the stock-overflow threshold during a set period. Support equipment. 9. The information on the production quantity is information calculated from the actual sales quantity against the planned sales quantity during the set period, and is the cumulative quantity of unachieved products and / or unachieved products weighted by product type. The support device for factory operation according to claim 1, which is an average achievement rate.