JPH0358231A - Membership function processing system - Google Patents

Abstract

PURPOSE:To manage in a lump membership functions corresponding to plural events by outputting the nearest discrimination value to an event value for making a ratio value obtained from a specific function maximum with regard to an input value, and a ratio value determined from this function by inputting the input value and the discrimination value. CONSTITUTION:At the time of outputting a prescribed event and a ratio for showing a degree of this event with respect to an input value, discrimination values 10 are allocated, respectively from a prescribed event value at every necessary event. Subsequently, a three-dimensional function 11 defined by the curved surface consisting of a locus of a straight line for connecting the corresponding points between closed curves consisting of a segment for connecting each point of every coordinate group, of two coordinate groups consisting of necessary points, respectively in which an input value, a ratio value and a discrimination value become three- dimensional coordinage values is provided so as to designate two different events. A function arithmetic part 12 outputs the nearest discrimination value 10 to an event value for making a ratio value obtained from this function with regard to the input value maximum, and a ratio value determined from this function by inputting the input value and the discrimination value, with respect to this input value. In such a way, membership functions corresponding to plural events can be managed in a lump.

Description

[Detailed Description of the Invention] [4 Summary] The purpose of the present invention is to provide a membership function processing method that allows membership functions corresponding to a plurality of events to be managed all at once, with regard to the processing of Menhasimbu functions used in so-called fuzzy operations. Then, when outputting a predetermined event and a ratio representing the degree of the event for the input value, each of the required events is assigned an identification value from the predetermined event value, and two different events are As specified, consists of a line segment connecting each point of each coordinate group of two sets of coordinate groups each consisting of the required points, each of which has the input value, the ratio value, and the identification value as the three-dimensional tower value. A three-dimensional function defined by a tilI surface consisting of a straight line trajectory connecting corresponding points between closed curves is provided, and the identification value closest to the event value that maximizes the ratio value obtained from the function for the input value is provided. and the ratio value determined from the function by inputting the input value and the identification value.

[Industrial application field]

The present invention relates to a second method for processing membership functions used in so-called fuzzy operations.

As is well known, the membership function is a function that calculates a value for fuzzy calculation, which indicates the degree of an event in a certain event classification representing a process state, etc., from an input value such as a measured value by a sensor.

[Conventional technology]

Fuzzy ( F
The u z z y ) calculation is also adopted in expert systems that diagnose process 4J[ItJi, etc. in the field of process processing.

In that case, for example, the temperature values measured by an appropriate sensor are categorized into several levels, from ``very low,'' ``slightly low,'' to ``very high,'' and for a certain temperature value, for example, ``80% A membership function is used to convert it into a fuzzy/11 arithmetic value such as ``slightly low.''

That is, as shown in Fig. 4, when the system is equipped with a function calculation unit l and an input value such as temperature and a function name are input, the function value of the function name is expressed as a value from 0 to 1.0, for example. outputs the ratio in which the function is applied, and represents the degree in the event frequency corresponding to that function.

For this purpose, the function calculation unit 1 has a function definition table 2, and according to definition information that is manually entered separately in advance, the function name and coordinate values (xH,
y. ) and approximated by a line segment connecting those points +
The input value is processed as described above using the th line as a specified function.

As shown in the temperature example above, there are generally many event classifications for input values in the l category, and one function is required for each classification. For example, when it comes to temperature, it is necessary to classify it into five categories such as ■ Very low ■ Slightly low ■ Normal ■ Slightly high ■ Very high, etc. As an example of the simplest form for each, we can use triangular functions to correspond to each other. Third
As shown in figure (a), each of the five functions has 2
~Specify and define three points.

When calculating a value for a certain input value, for example, in the case of the input value shown by a in Figure 3 (a), it is normal at about 770% due to the relationship of ■, which is the largest value among each function value.'' shall be.

[Problem to be solved by the invention]

In this way, it is necessary to provide multiple functions for one sensor, etc., but since the system is generally designed to process sensor input from multiple categories and perform comprehensive diagnosis, etc., management is required for this purpose. There is a problem in that the number of membership functions becomes extremely large, and the work of modifying the functions for optimization adjustment of the system becomes complicated.

An object of the present invention is to provide a membership function processing method that allows membership functions corresponding to a plurality of events to be collectively managed.

[Means to solve the problem]

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

The figure shows the structure of the membership function processing method, and when outputting a predetermined event and a ratio representing the degree of the event for an input value, each required event is identified from the predetermined event value. 4ialO, and the coordinate groups of two sets of coordinates each consisting of a required point that three-dimensionally coordinates the input value, the ratio value, and the identification value so as to specify the two different events. Closed +H consisting of line segments connecting each point of
A three-dimensional function 11 defined by a curved surface formed by the locus of straight lines connecting corresponding points between the I lines is provided.

The function calculation unit 12 calculates, for the input value, the identification value 10 closest to the event value that maximizes the ratio value obtained from the function for the input value, and manually calculates the input value and the identification value. The ratio value determined from the function is output.

[Operation] With the above processing method, there is no need to create and manage functions for each h of events in the same category, and one function for each category can be defined using a relatively small number of coordinate points. can replace the conventional multiple function definitions.

[Example] The function calculation unit l2 in FIG. 1 functions two sets of three-dimensional coordinate groups, a group of coordinates (XIi+Vli+ZI) and a group of coordinates (Xzi, y2i+*2), according to definition information input separately in advance. It is held in definition table 11. Here, the X and y coordinates are made to correspond to the input value and the ratio value, as in the conventional case. Further, the two coordinates are made to correspond to event values, and specific event values are assigned to each required event and held in the identification value table 10 as event identification values.

The above two sets of coordinates are two different coordinates selected from multiple events.
Each event consists of coordinate values that define a membership function for events z1 and z2.

The function calculation unit 12 uses these two coordinate groups to define a three-dimensional Yikyu defined by the IllJ plane, which is a locus of a straight line connecting corresponding points between closed curves made of line segments connecting each point of each coordinate group. Assuming that the point on the +To plane is a specified function, the input clay is processed as follows.

That is, as shown in FIG. 2, for a certain input value X in processing step 20, for example, input value X and ratio value y=1.0
The function value Z determined from is determined, and in processing step 21, the identification value 2 closest to Z is determined. is determined using identification value table 10. Next, in processing step 22, input value X and event value 2 are input. Find the function value y determined from k, and process [11j Z with step 23. and y are output.

FIG. 3(b) is a diagram explaining a three-dimensional membership function that replaces the five membership functions illustrated in FIG. 3(a). For example, the functions of Each x-y coordinate used in the definition to define those functions and the identification value Z assigned to "very low" and "very high" events. 1 and ZoS to the above 21 and z2
Define a three-dimensional function using the set of coordinates.

Also, Z on the event value axis. Find 204, 203, and 204 to divide the interval from 1 to zos into four equal parts, for example,
These are the identification values for "slightly low", "normal 1" and "slightly high" events, respectively.

As a result, the function calculation unit 12 calculates this function by assuming that the triangular prism specified by the conventional ■ and ■ functions has a cross section parallel to the Process the input value as described above, and 2
Outputs the difference between 01 and zos and the ratio value y.

By doing this, as is clear from Fig. 3, conventionally two or three definition points are required for each of the five functions,
A total of 13 points had to be managed, but by creating one three-dimensional function, the function can now be defined using only six points.

Note that the above explanation uses an example of defining a function with two cross sections, but if necessary, the function corresponding to the l category can be defined by dividing it into two or more intervals each defined with two cross sections. In that case as well, each section can be handled in the same way as in the above example.

〔Effect of the invention〕

As is clear from the above explanation, according to the proposed Iyl, in the processing of membership functions used in so-called fuzzy operations in computers, member simple functions corresponding to multiple events in the same category are collectively processed in a relatively small number. This has a significant industrial effect of improving the efficiency of membership function management and processing.

[Brief explanation of drawings]

Fig. 1 is a block diagram showing the configuration of the present invention, Fig. 2 is a flowchart of the processing of the invention, Fig. 3 is an explanatory diagram of the Menhashinbu function, and Fig. 4 is a block diagram showing an example of a conventional construction. In the figure, 1 and l2 are function operation parts, 2 and 11 are function constant tables, and 10
20-23 are a block diagram showing the structure of the processing step tree invention. Fig. 1 is a flowchart of the processing of the present invention. Fig. 2 is an explanatory diagram of the Membang Knob function. Fig. 3 is a block diagram showing an example of the conventional structure.

Claims (1)

[Claims] When outputting a predetermined event and a ratio representing the degree of the event for an input value, each of the required events is assigned an identification value from the predetermined event value (10). , In order to designate two different events, each coordinate group of two sets of coordinates each consisting of required points with the input value, the ratio value, and the identification value as three-dimensional coordinate values. A three-dimensional function (11) defined by a curved surface consisting of the locus of a straight line connecting corresponding points between closed curves consisting of line segments connecting points is provided, and the ratio value obtained from the function (11) for the input value is calculated. It is characterized by being configured to output (12) the discrimination value closest to the event value to be made the highest, and the ratio value determined from the function using the input value and the discrimination value as input. membership function processing method.