JPH08153002A - Fuzzy inference device and fuzzy inference method - Google Patents

Abstract

PURPOSE: To easily constitute a fuzzy inference device in accordance with change in the type of problem to be treated and observation data and to easily correct knowledge on the addition of various procedures required for inference. CONSTITUTION: The method required for fuzzy inference in constituted of four types of an input type 4, a translation type 5, a synthesis type 6 and an interpretation type 7. They are adaptively selected by an inference execution management part 3. In the method of the input type 4, the procedure for allocating a basic probablity to the focus element of D-S logic by using fuzzy inference is described. In the methods of the translation type 5, the synthesis type 6 and the interpretation type 7, the inference procedure based on D-S theory is described.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION The present invention uses fuzzy logic in inference to integrate input data with uncertainty using the DS (Dempster-Shafer) theory.
The present invention relates to a fuzzy inference apparatus and a fuzzy inference method for converting given input data into input values for inference based on the DS theory.

[0002]

2. Description of the Related Art Input information and diagnosis / control knowledge for a diagnosis / control system are not always reliable knowledge. Therefore, it is often necessary to express uncertain information and knowledge. Further, in a diagnostic / control system, it is often necessary to make inferences based on observation data with uncertainty and integrate a plurality of data to draw a conclusion. Among the systems that integrate these plural observation data and perform diagnosis and control, there are systems that use the DS theory. The basic concept of the DS theory will be described below.

[DS Theory] (Basic Concept of DS Theory) In the 1960s, Dempster proposed Ba
In order to deal with subjective probabilities that are difficult to handle with yes probabilities, we have proposed the concepts of lower probability and upper probability. In addition, Shafer uses this Dempster
Expanding the theory of, the lower bound probability and the upper bound probability
In other words, we defined them as ef function and plausibility based on basic probability.

In the D-S theory, a general set having possible events as elements is defined as a universal set (frame of discernment). A major feature of the DS theory is that a quantity called basic probability is assigned to a subset of the universal set θ. Below are the basic concepts in the DS theory: basic probability, focus element, upper bound, lower bound probability, belief interv
Explain al.

1) Basic probability m (Ai) (i = 1,2, ...)
Takes a value of [0,1] and satisfies the following conditions.

M (φ) = 0 (1) (φ: empty set)

[Equation 1] (Θ: universal set, Ai (i = 1,2, ...): Subset of θ) 2) The focal element is Ai satisfying the following condition.

M (Ai)> 0 (3) 3) Lower bound probability

[Equation 2] 4) Upper bound probability

(Equation 3) The confidence of A is Bel (A) at the minimum and 1-Bel at the maximum.
(Ac).

Below, these basic concepts will be explained using the example shown in FIG.

In the following example, the possible elements are a1 to a6, the entire set is θ, and its subset A is
1 to A6.

Θ = {a1, a2, a3, a4, a5, a6} A1 = {a2} A2 = {a4} A3 = {a1, a2, a3} A4 = {a4, a5} A5 = {a2, a3 , A4, a5} A6 = {a1, a2, a3, a4, a5, a6} It is assumed that a basic probability m (Ai) is assigned to these subsets as shown in FIG. 22 below.

In the following, a set of a subset (focus element) to which the basic probability is assigned and the basic probability will be represented by the following list.

(((A2) 0.1) ((a4) 0.2) ((a1, a2, a3) 0.1) (7) ((a4, a5) 0.3) ((a2, a3, a4, a5) 0.1) ( (A1, a2, a3, a4, a5, a6) 0.2)) At this time, for example, the subset A4 of FIG. 22 = {a4, a5
} The lower and upper bound probabilities and belief intervals are as follows.

・ Lower bound probability

[Equation 4] ・ Upper bound probability

(Equation 5) ・ Belief interval

(Equation 6) The probability that A4, that is, {a4, a5} is at least 0.
5, which indicates that it may increase up to 0.8 in some cases.

(Dempster's Combining Rule) This rule is a method of integrating basic probabilities inferred from independent evidence.

[0015]

(Equation 7) m1, m2: basic probabilities obtained based on mutually independent evidences E1 i, E2 j (i, j = 0,1,2, ...): focus element of m1, m2 Dempster's connection rule of equation (8) By combining the respective basic probabilities, a new basic probability is obtained.

Further, when the combining rule is applied, a basic probability may be assigned to the empty set. However, as a prerequisite for the basic probability, the basic probability of the empty set is 0.
Therefore, the amount is inconsistent. Therefore, it is necessary to divide other basic probabilities by the amount obtained by subtracting the inconsistent amount from the whole and normalize it. The operation is performed with the denominator of the above formula.

FIG. 22 below shows two independent evidences E1 = ((E1 1 0.9) (E1 2 0.1) = (((a2) 0.9) (9) ((a1 , A2, a3, a4, a5, a6) 0.1)) E2 = ((E2 1 0.8) (E2 2 0.2) = (((a1, a2, a3) 0.8) (10) ((a2, a3, a4, a5) 0.2)) are combined, where, as mentioned above, the list (9) shows that the focus element E1 1 of the evidence E1 has a basic probability of 0.9.
Shows that the focus element E1 2 is assigned a basic probability of 0.1.

According to FIG. 23, the combination result of two independent evidences E1 and E2 is (((a2) 0.72) ((a1, a2, a3) 0.08) ((a2) 0.18) (11) ((a2 , A3, a4, a5) 0.02)). Here, the first list ((a2) 0.72) and the third list ((a2) 0.18) of this list (11) have the same focus element (a2), so both are combined. Therefore, the final integration result is (((a2) 0.90) ((a1, a2, a3) 0.08) (12) ((a2, a3, a4, a5) 0.02)).

Further, in recent years, fuzzy logic has been widely used in many knowledge processing systems to express ambiguity in meaning and definition of words. The fuzzy logic is briefly described below.

[Fuzzy Logic] In fuzzy logic, a membership function is used to express a concept that is difficult to express with a true or false binary value.

For example, the expression "child of normal grade" is ambiguous, and in order to strictly define this, the expression "child with an average score of 2.5 or more and less than 3.5" must be used. . However, the majority of expressions are ambiguous in everyday life, and the person who accepts them accepts the ambiguity as an implicit understanding. Now, expressing the concept of "ordinary grade child" using a membership function is as shown in FIG. The set represented by (a) is called a fuzzy set. From this graph, the degree of membership of a child with a score of 2.5 is 0.5, which is 3.0.
Gradually approaching 1.0, moving away from 1.0 again, the degree of membership of 3.5 children is 0.5. That is, the value in the exact expression depends on whether it belongs to the expression, 0 or
It was either one, but in fuzzy sets its value
It is defined by an arbitrary degree of membership from 0 to 1. That is,
In the graph of (a) of FIG. 24, a child who has a complete expression of "ordinary grade child" has a score of 3, and the range is wide. When a strict expression such as "a child whose average score is 2.5 or more and less than 3.5" is used, it can be expressed as shown in FIG. That is, it is also possible to perform conventional strict expression in the framework of fuzzy sets. When distinguishing a set having an exact expression as shown in FIG. 24B from a normal fuzzy set, this set is called a "crisp set". Such fuzzy logic is widely used in fields where more human and ambiguous judgments and decisions are required.

By the way, the observation data given to the system using the DS inference is often in a numerical format such as "amount of rainfall 10 mm". On the other hand, in the DS inference, the observed data with uncertainty such as "the possibility of light rain is 90%, and the possibility of light rain or moderate rain is 10%." As required. In this way, there is a large difference between the given observation data and the input value of the DS inference, and the observation data is converted into the input value of the DS inference by using some "fuzzy judgment knowledge". Pre-processing for format conversion is required. This data conversion, which is a preprocessing of the DS inference, is called "assignment of basic probability to focus element".

However, in the conventional "assignment of basic probability to focus element", human beings subjectively judge observation data to create a basic probability value, or "if total rainfall is 10 mm, The chance of light rain is 90%. "It was assigned by memorizing the crisp knowledge in the computer. Therefore, the computer said, "If the total rainfall is 10 mm, the probability of light rain is 90%,
The chance of light or moderate rain is 10%. It was not possible to assign the basic probability by effectively utilizing the human's "fuzzy judgment knowledge" such as ".

Therefore, for example, in Japanese Unexamined Patent Publication No. 2-278431, only the antecedent part of the production rule required for inference is described using fuzzy sets to include ambiguity, and the inference result is derived using the Dempster's associative rule. A method has been proposed.

However, in such a method, for example, when the problem to be handled or the type of observation data is changed,
It is necessary to reconfigure the device each time. In addition, it is not always easy to modify knowledge such as adding various procedures necessary for inference.

[0026]

SUMMARY OF THE INVENTION The present invention has been made to solve the above-mentioned problems, and the device can be easily configured according to the problem to be handled and the type of observation data, and the reasoning It is an object of the present invention to provide an ambiguous inference apparatus and an ambiguous inference method in which it is easy to modify knowledge such as addition of various necessary procedures.

Another object of the present invention is to provide a fuzzy reasoning apparatus and a fuzzy reasoning method which can process fuzzy reasoning more quickly.

[0028]

In order to achieve the above object, the fuzzy inference apparatus of the present invention uses a fuzzy inference to convert observation data into a first data list with confidence. Means, a second data conversion means for converting the first data list with the certainty factor into a second data list with the certainty factor based on the conversion rule list, and a plurality of types based on the Dempster combination rule Based on the Dempster-Shafer probability theory, and a data unifying means for unifying a plurality of types of the second data list with the certainty factor corresponding to the observation data of the third data list with the certainty factor. Data calculating means for calculating a fourth data list with upper and lower bound probabilities from the data list, and the first data converting means, the second data converting means, and the data integrating means. And a starting order determining means for determining the starting order of the data calculating means, and the first data converting means, the second data converting means, the data integrating means or the data calculating means according to the determined starting order. And an execution management unit for sequentially activating and executing.

Further, the fuzzy reasoning apparatus of the present invention, based on the fuzzy reasoning, converts the observation data into a first data list with confidence, and a first data conversion means, and based on the conversion rule list, the confidence. Second data conversion means for converting the first data list with a certainty into the second data list with a certainty factor, and a plurality of types of the certainty corresponding to a plurality of types of observation data based on the Dempster's combining rule A data integration means for integrating the second data list with a degree into the third data list with a certainty degree, and Dempster-Shafer
Data calculation means for calculating a fourth data list with an upper bound probability and a lower bound probability from the third data list based on the probability theory of, a first data conversion means, a second data conversion means, A starting order determining means for determining the starting order of the data integrating means and the data calculating means, a membership function required for the fuzzy inference, a template of the first data list with the certainty factor, and the conversion rule list. Input means for inputting, first data converting means according to the determined starting order, second data converting means, data integrating means, data calculating means, the input membership function, First with confidence
Means for preliminarily associating the template of the data list of the above with the conversion rule list before executing the fuzzy inference, and the template of the first data list with the associated starting order, membership function, and certainty factor. According to the form and the conversion rule list, the first data conversion means, the second data conversion means, the data integration means or the execution management means for sequentially activating and executing the data calculation means.

The fuzzy reasoning apparatus of the present invention uses the fuzzy reasoning to convert the observation data into a first data list with confidence, and a first data conversion means based on the conversion rule list. Second data conversion means for converting the first data list into a second data list with certainty factor, and a plurality of types of the certainty factor corresponding to a plurality of types of observation data based on Dempster's combining rule A second data list of the third data list with certainty factor, and a fourth data with upper and lower bound probabilities from the third data list based on the Dempster-Shafer probability theory. Data calculating means for calculating the data list of the first data converting means, the first data converting means, the second data converting means, the data integrating means, and the data calculating means during execution of fuzzy inference. It comprises a determining means for sequentially deciding a means to be activated and executed among the means, and an execution managing means for activating and executing the determined means.

The fuzzy inference apparatus of the present invention uses the fuzzy inference to convert the observation data into the first data list with confidence, and the first data conversion means, and based on the conversion rule list, adds the confidence. Second data conversion means for converting the first data list into a second data list with certainty factor, and a plurality of types of the certainty factor corresponding to a plurality of types of observation data based on Dempster's combining rule A second data list of the third data list with certainty factor, and a fourth data with upper and lower bound probabilities from the third data list based on the Dempster-Shafer probability theory. The data calculation means for calculating the data list, the membership function required for the fuzzy inference, the template of the first data list with the certainty factor, and the conversion rule list. Input means for inputting and sequentially executing the means to be activated and executed among the first data conversion means, the second data conversion means, the data integration means and the data calculation means during execution of fuzzy inference. While associating the determining means for determining, and the determined means with the input membership function, template or conversion rule list of the first data list with certainty factor,
And an execution management unit that activates and executes the unit.

The fuzzy inference apparatus of the present invention uses the fuzzy inference to convert the observation data into a first data list with confidence, and a first data conversion means, and based on the conversion rule list, adds the confidence. Second data conversion means for converting the first data list into a second data list with certainty factor, and a plurality of types of the certainty factor corresponding to a plurality of types of observation data based on Dempster's combining rule A second data list of the third data list with certainty factor, and a fourth data with upper and lower bound probabilities from the third data list based on the Dempster-Shafer probability theory. Data calculating means for calculating the data list, and the first data converting means, the second data converting means, the data integrating means or the data calculating means according to the contents to be inferred. And a means for appropriately combining the means.

The fuzzy reasoning apparatus of the present invention uses the fuzzy reasoning to convert the observation data into a first data list with confidence, and a first data conversion means, and based on the conversion rule list, adds the confidence. Second data conversion means for converting the first data list into a second data list with certainty factor, and a plurality of types of the certainty factor corresponding to a plurality of types of observation data based on Dempster's combining rule A second data list of the third data list with certainty factor, and a fourth data with upper and lower bound probabilities from the third data list based on the Dempster-Shafer probability theory. The data calculation means that calculates the data list of the above and the whole inference process that integrates multiple types of observation data are divided into modules, and the data flow that links each module with a relational link. -Input means for inputting a diagram, and means for appropriately combining the first data conversion means, the second data conversion means, the data integration means or the data calculation means according to the input data flow diagram To do.

In the above fuzzy reasoning device,
At least a plurality of observation data composed of meteorological data and map data of a predetermined area, and an evaluation value of the corresponding area may be integratedly calculated as a regional analysis result based on knowledge previously created by humans. .

The fuzzy inference method of the present invention uses a fuzzy inference to convert observation data into a first data list with confidence, and a first rule with confidence based on the conversion rule list. Second procedure for converting the data list of the data into the second data list with certainty, and Demp
a third procedure for integrating a plurality of types of the second data list with certainty corresponding to a plurality of types of observation data into a third data list with certainty based on the ster's combining rule; and Dempster -Based on Shafer's theory of probability, a fourth procedure for calculating a fourth data list with upper and lower bound probabilities from the third data list is prepared as a component and prepared in advance, and contents to be inferred It is characterized in that these first to fourth procedures are appropriately combined according to the above.

Further, the fuzzy inference method of the present invention uses the fuzzy inference to convert the observation data into a first data list with certainty factor and a conversion rule list based on the first procedure. A second procedure for converting the first data list into a second data list with certainty factor, and a plurality of types of the certainty factor-based first data list corresponding to a plurality of types of observation data based on the Dempster coupling rule. A third procedure for integrating the second data list into a third data list with confidence and a fourth procedure with upper and lower bound probabilities from the third data list based on the Dempster-Shafer probability theory. The fourth procedure for calculating the data list is prepared as a component and prepared in advance, the activation order of these first to fourth procedures is determined according to the content to be inferred, and according to the determined activation order, It is characterized in that these first to fourth procedures are sequentially activated to execute inference.

The fuzzy inference method of the present invention comprises a first procedure for converting observation data into a first data list with certainty factor based on fuzzy inference, and a first procedure with certainty factor based on a conversion rule list. Second procedure for converting the data list of the data into the second data list with certainty, and Demp
A fourth procedure for calculating a fourth data list with upper and lower bound probabilities from the third data list based on ster's coupling rule theory is prepared in advance as a component.
The invocation order of these first to fourth procedures is determined according to the content to be inferred, and the inference is executed by sequentially invoking these first to fourth procedures in accordance with the determined invocation order. .

The fuzzy inference method of the present invention is based on a fuzzy inference to convert the observation data into a first data list with certainty factor, and a first rule with certainty factor based on the conversion rule list. Second procedure for converting the data list of the data into the second data list with certainty, and Demp
a third procedure for integrating a plurality of types of the second data list with certainty corresponding to a plurality of types of observation data into a third data list with certainty based on the ster's combining rule; and Dempster -Based on Shafer's probability theory, a fourth procedure for calculating a fourth data list with upper and lower bound probabilities from the third data list is prepared as a component, and fuzzy inference is executed. It is characterized in that among the first to fourth procedures, the procedure to be activated and executed is sequentially determined, and the determined procedure is activated and executed.

[0039]

In the present invention, the fuzzy inference processing is divided into the above-mentioned four procedures (including means corresponding thereto), and the fuzzy inference is executed by combining these procedures according to the content of the inference. Therefore, it is possible to easily configure the device according to the problem to be handled and the type of observation data, and it is easy to modify the knowledge such as adding various procedures necessary for inference.

Further, the activation order of these procedures is determined according to the contents to be inferred, and the activation order is determined according to the determined activation order.
Since these procedures are sequentially activated and the inference is executed, the process during the inference execution can be reduced to only the calculation of the certainty factor and the update process, and the inference can be speeded up.

Further, during execution of fuzzy inference, the procedure to be activated among these procedures is sequentially determined, and the determined procedure is activated and executed. The processing of the observed data can be advanced before the next observation data is input,
For example, when the observation data is input intermittently instead of continuously, the inference process can be performed more quickly.

[0042]

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 is a diagram showing the configuration of a fuzzy reasoning apparatus according to an embodiment of the present invention. In this embodiment,
An example of calculating the mobility cost from two observation data of "rainfall" and "road congestion" will be described. Here, the mobility cost is an index indicating the ease of movement of a vehicle at a certain point.

The ambiguous inference apparatus shown in FIG. 1 includes an inference procedure generation unit 1 for determining the activation order of each method based on a data flow diagram 11, an inference preprocessing execution unit 2 for initializing inference, and a corresponding unit. The inference execution management unit 3 which issues a start instruction to the method to be executed, and the method 4 in which an arithmetic expression for performing fuzzy inference is described as procedural knowledge
7, a fuzzy inference execution unit 8 for converting an input environment value into an input value of the DS theory, and D, which uses the DS theory to integrate information including uncertainty and execute fuzzy inference −
The S inference execution unit 9 is included.

Methods 4 to 7 are input type 4, translation type 5,
It is composed of four types of integrated type 6 and interpretive type 7, and is selected by the inference execution management unit 3. The input type method 4 describes a procedure for assigning a basic probability to the focus element of the DS theory by using fuzzy inference, and the DS type theory is used for the translation type 5, integrated type 6 and interpretive type 7 methods. The inference procedure based on is described. That is, the method of translation type 5 converts the data list with certainty factor based on the conversion rule list. The method of integrated type 6 is Dempst
Based on er's combining rule, data lists with certainty factors corresponding to multiple types of observation data are integrated. The method of the interpretation type 7 calculates a data list with upper and lower bound probabilities from the integrated data list based on the DS probability theory.

The object information 10, which is a storage area of the fuzzy inference apparatus, includes an attribute value of each object, an inference procedure created by the inference procedure generation unit 1, a data flow diagram 11 representing a relationship between objects, and an input. A universal set 12 used in the processing of objects, a conversion rule 13 required in the processing of translated objects,
A membership function 14 used for fuzzy reasoning,
Environment value 1 that stores the observed data that is the source of the inference
5 and the data of the inference intermediate result calculated by the inference execution management unit 3 are stored. Then, the inference final result calculated by the inference execution management unit 3 is stored in the inference result 16.

The data flow diagram 11 is, for example, as shown in FIG. 2, in which the whole inference process for integrating two observation data is divided into modules and each module is connected by a relational link. Incidentally, the data flow diagram 11 shown in FIG. 2 shows the inference process of calculating the activation cost from the two observation data of “rainfall” and “road congestion”. This inference process is broken down into four processes: input, translation, integration, and interpretation. Inside the object information 10, each of the decomposed processes is represented as one object. FIG. 3 shows the contents of each of these objects. Each object shown in the same figure has attributes O11, O12, O21, O22, O31, O32, O41, O42, O for storing a processing result list of the object and information used in the processing process.
51, O61, O62, O71 and methods O13, O23, O33, O43, O52, O63, O72 in which the processing procedures are described
Is stored.

In the fuzzy reasoning apparatus of this embodiment, this
Data flow diagram 11, observation data and DS
A user has created in advance a membership function as shown in FIG. 4, for example, which expresses an ambiguous human judgment regarding correspondence with the focus element of the theory.

Next, the operation of the ambiguous inference device configured as described above will be described.

FIG. 5 shows the procedure of the overall processing in this fuzzy reasoning apparatus. That is, the processing in the present embodiment can be roughly divided into three: the initial setting (step S1), the method operation (steps S2 to S7), and the post-processing (step S8). 6 to 9 show the detailed procedure of the initial setting, FIGS. 10 to 13 show the detailed procedure of each method, and FIG. 14 shows the detailed procedure of the fuzzy inference execution part in the method 1.

First, initial setting is performed (step S1 in FIG. 5).

The initialization is divided into three processes: determination of inference procedure (step T1 in FIG. 6), reading of data (step T2 in FIG. 6), and initialization of list (step T3 in FIG. 6).

First, the inference procedure generator 1 determines an inference procedure based on the data flow diagram 11 (step T1 in FIG. 6). In the data flow diagram of this embodiment, as shown in FIG. Input (rainfall) 2. Translation (rainfall) 3. Input (road congestion) 4. Translation (road congestion) (13) 5. Integration 6. Translation (integration result) 7. Decide the inference procedure like interpretation. Then, the determination result is stored in the object information 10 in a list format (steps U1 to U3 in FIG. 7). That is, the relationship link information of the data flow diagram of the object as shown in FIG. 2 is acquired (step U1 in FIG. 7), and the processing order list is created based on the acquired relationship link information (step U in FIG. 7).
2) The created processing order list is stored in the object information 10 (step U3 in FIG. 7).

Second, in the inference preprocessing execution unit 2, each of the data (universal set data 12, conversion rule data 13, membership function data 14) required in the inference process is read and stored in the object information 10 ( FIG. 8 steps V1 to V3).

The universal set data 12 means data on a whole set having the event used in each observation data as an element. In the present embodiment, the universal set data of "rainfall" is a set having three events of (LIGHT-RAIN MODERATE-RAIN HEAVY-RAIN) (14) as elements, and in the "road congestion state", (LIGHT-CROWDED MODERATE-CROWDED HEAVY-CROWDED) (15).

The conversion rule data 13 is (LIGHT-RAIN MODERATE-RAIN
This is a rule for converting an event such as HEAVY-RAIN) into a mobility event (FAST NORMAL SLOW), and is used in a translation object. For example, a translation (rainfall) object (O2 in Figure 3
The conversion rule data of () is ((LIGHT-RAIN FAST) (MODERATE-RAIN NORMAL) (16) (HEAVY-RAIN SLOW)), and the conversion rule of the translation (road congestion status) object (O4 in FIG. 3). The data is ((LIGHT-CROWDED FAST) (MODERATE-CROWDED NORMAL) (17) (HEAVY-CROWDED SLOW)). For example, in the list (16), (LIGHT-RAIN FAST) reads "Convert LIGHT-RAIN to FAST."
Represents the conversion rule. Similarly, the conversion rule data of the translation (integration result) object (O6 in FIG. 3) is also ((FAST & FAST FAST) (FAST & NORMAL NORMAL) (NORMAL & NORMAL NORMAL) (NORMAL & FAST NORMAL) (FAST & SLOW SLOW) (18) & (SLOW). NORMAL & SLOW SLOW) (SLOW & NORMAL SLOW) (SLOW & SLOW SLOW)). These conversion rule lists are the attributes (conversion rules) of the corresponding objects (O22, O42,
It is stored in O62).

The membership function data 14 is data defining a membership function relating to each environmental factor. For example, the membership function data regarding “rainfall” is expressed in the following list format.

(((LIGHT-RAIN) (1.0 0 20 20 35)) ((LIGHT-RAIN MODERATE-RAIN) (1.0 15 32.5 32.5 50)) ((MODERATE-RAIN) (1.0 25 40 60 75)) ( (LIGHT-RAIN MODERATE-RAIN HEAVY- RAIN) (0.2 30 35 65 70)) ((MODERATE-RAIN HEAVY-RAIN) (1.0 50 67.5 67.5 85)) ((HEAVY-RAIN) (1.0 65 80 100 100)) ((HEAVY-RAIN HEAVY-RAIN) (0 0 0 0 0))) (19) Each list in this list (19) is ((focus element) (maximum value of degree of membership first definition point second definition) Points The third definition point and the fourth definition point)))) are used to represent the membership function as shown in Fig. 4. In addition, the first to fourth definition points indicate "total rainfall of rain within 12 hours", and the unit thereof is mm. Similarly, regarding the “road congestion situation”, (((LIGHT-CROWDED) (1.0 0 0 10 15)) ((LIGHT-CROWDED MODERATE-CROWDED) (1.0 5 22.5 22.5 40)) ((MODERATE-CROWDED) ( 1.0 15 30 50 65)) ((LIGHT-CROWDED MODERATE-CROWDED HEAVY-CROWDED) (0.2 20 25 55 60)) ((MODERATE-CROWDED HEAVY-CROWDED) (1.0 40 57.5 57.5 75)) ((HEAVY-CROWDED (1.0 55 70 100 100)) ((HEAVY-CROWDED HEAVY-CROWDED) (0 00 00 0)))) is defined as (20). The first to fourth definition points in the above list (20) indicate "time required to proceed 5 km", and the unit is minutes. These membership function list (19) (or 20) is the attribute (membership function) of the corresponding input object (O12, O in FIG. 3).
32).

Third, the inference preprocessing execution unit 2 initializes the list for each object of input, translation, integration, and interpretation. (FIG. 6, step T3).

First, input objects (O1, O in FIG. 3)
3) Initialize the list. That is, the focus element list in the DS theory is generated based on the universal set data (14) (or (15)) previously read (step W1 in FIG. 9). Here, the focus element list to be generated is a list including all subsets (excluding the empty set) of the events of the universal set list (14). Below, the focus element list (21) regarding “rainfall” and the focus element list (22) regarding “road congestion” are shown. (((LIGHT-RAIN) 0) ((LIGHT-RAIN MODERATE-RAIN) 0) ((MODERATE-RAIN) 0) ((LIGHT-RAIN MODERATE-RAIN HEAVY-RAIN) 0) ((MODERATE-RAIN HEAVY-RAIN ) 0) ((HEAVY-RAIN) 0) ((LIGHT-RAIN HEAVY-RAIN) 0)) (21) (((LIGHT-CROWDED) 0) ((LIGHT-CROWDED MODERATE-CROWDED) 0) ((MODERATE- CROWDED) 0) ((LIGHT-CROWDED MODERATE-CROWDED HEAVY-CROWDED) 0) ((MODERATE-CROWDED HEAVY-CROWDED) 0) (( HEAVY-CROWDED) 0) ((LIGHT-CROWDED HEAVY-CROWDED) 0)) (22) The numerical value 0 at the end of each list means the possibility of each event (each seismic intensity). There is. For example,
((LIGHT-CROWDED) 0 of list (22)
) Is interpreted as “There is a 0% chance that the road at that mesh point will have a slight traffic jam.”

Translation objects (O2, O4, O in FIG. 3)
The list of 6) is initialized in the same way as the input object. That is, the previously read conversion rule data (1
6) (or (17)), the list is generated (step W2 in FIG. 9).

The list generated by this translation object is one in which the atom has been converted into one that means mobility, and is the same for both "rainfall" and "road congestion" (((( FAST) 0) ((FAST NORMAL) 0) ((NORMAL) 0) ((FAST NORMAL SLOW) 0) (23) ((NORMAL SLOW) 0) ((SLOW) 0) ((FAST SLOW) 0)) is there.

Initialization of the integrated object (O5 in FIG. 3) is performed using the list (23) initialized by the two upper objects of "rainfall" and "road congestion". That is, the list (24) generated by the integrated object is composed of 49 lists including the atoms connecting the atoms of the two upper-level object lists (23) with & as shown below (Fig. 9 steps W
3). (((FAST & FAST) 0) ((FAST & FAST FAST & NORMAL) 0) ((FAST & NORMAL) 0) ((FAST & FAST FAST & NORMAL FAST & SLOW) 0) ((FAST & NORMAL FAST & SLOW) 0) (FAST & NORMAL FAST & SLOW) 0) FAST & FAST NORMAL & FAST) 0) (24) ... ((SLOW & SLOW NORMAL & SLOW) 0) 0) ((SLOW & LOW SLOW & NORMAL) 0) ((Initializing the list in SLOW & SLOW) 0) interpretation object (O7 in FIG. 3) to generate the following list (25) (FIG. 9 step W4).

(((FAST) 0 0) ((NORMAL) 0 0) (25) ((SLOW) 0 0)) In the interpretation object, the upper and lower bound probabilities of the DS theory are obtained, so the above list ( As shown in 25), 0 is set at two places so that upper and lower bound probabilities can be stored at the end of each list.Next, the inference execution management unit 3 selects a method to be activated. 10
According to the inference procedure (13) stored in, the method of the input (rainfall) object at the beginning of the inference procedure (O13 in FIG. 3) is followed by the method of the input (rainfall) object at the beginning of the inference procedure (FIG. 3). Select O13) (Fig. 5)
Step S2).

The method type of the input (rainfall amount) object is checked to confirm that it is the input type method 1 (step S3 in FIG. 5).

Then, method 1 is executed (step S4 in FIG. 5).

Method 1 starts with the focus element list (2
Initialize the basic probability of 1) to 0. This time, since the basic probability of the focus element list (21) is set to 0 at the time of generation, no processing is performed (step A1 in FIG. 10).

Next, the environment value 15 is read. "Rainfall"
Environment value is "total amount of rain that fell within 12 hours at the mesh point (unit: mm)", and 0 to 1
Given in 00 mm. In this embodiment, it is assumed that the “rainfall amount” is given as “17 mm” (FIG. 10, step A).
2).

Here, the membership function list (19) read previously is acquired (step A in FIG. 10).
3).

With the above steps, the preparations are completed and the fuzzy inference is executed (step A4 in FIG. 10).

In the execution of fuzzy inference, first, the environment value is read into the membership function. This state is shown in FIG. 15 (step B1 in FIG. 14).

In FIG. 15, the input environment value is indicated by a thick line. This thick line indicates (LIGHT-RAIN) and (LIGHT-RAIN)
MODELATE-RAIN), and the corresponding degree of membership of the intersection is (LIGHT-RAIN) 1.0, (LIGHT-RAIN MODERATE-RAI
It can be seen that N) is 0.11. Therefore, ((LIGHT-RAIN) 1.0) ((LIGHT-RAIN MODERATE-RAI
N) 0.11), the degree of belonging is temporarily assigned to each focus element (step B2 in FIG. 14).

Next, the degree of membership of each focus element is divided by the sum of the degrees of membership of focus elements, the degree of membership temporarily assigned to each focus element is normalized, and a basic probability is assigned. That is, regarding (LIGHT-RAIN),

(Equation 8) Next, (LIGHT-RAIN MODERATE-
Regarding RAIN),

[Equation 9] (Step B3 in FIG. 14).

As described above, (LIGHT-RAIN)
At 0.9, (LIGHT-RAINMODERATE-R
The basic probability of 0.1 is assigned to AIN), and the basic probability of the focus element list (21) is (((LIGHT-RAIN) 0.9) ((LIGHT-RAIN MODERATE-RAIN) 0.1) ((MODERATE-RAIN) 0). ((LIGHT-RAIN MODERAIN HEAVY-RAIN) 0) ((MODERATE-RAIN HEAVY-RAIN) 0) ((HEAVY-RAIN) 0) ((LIGHT-RAIN HEAVY-RAIN) 0)) (28) (Step A5 in FIG. 10).

Next, the list (28) generated by the processing of the input object using the fuzzy inference is stored in the attribute (processing result) of the own object as a processing result list (step A6 in FIG. 10).

Here, the process returns to the method selection, and the method (O23 in FIG. 3) of the next translation (rainfall) object is selected according to the inference procedure (13) (step S2 in FIG. 5).

Method of Translation (Amount of Rainfall) Object This is a translation type method 2 (step S3 in FIG. 5).

Then, the method 2 is executed (step S4 in FIG. 5).

The method 2 is initialized so that the certainty factor of the list (23) initially generated at the time of initialization (step S1 in FIG. 5) becomes zero. Similar to method 1, this time,
Since the certainty factor of this list (23) is set to 0 at the time of generation, no processing is performed (step X in FIG. 11).
1).

Next, the conversion rule list (1 that was previously read and stored in the attribute (conversion rule) of the own object)
6) is acquired (step X2 in FIG. 11).

Further, the basic probability of each focus element is acquired from the list of the attributes (process results) of the upper object. The upper object of the translation (rainfall) object is an input (rainfall) object (O1 in FIG. 3) as shown in the data flow diagram of FIG. That is, the basic probability acquired here is ((LIGHT-RAIN) 0.9) ((LIGHT-RAIN MODERATE-RAI) of the focus element list (28) of the input (rainfall) object.
N) 0.1) (step X3 in FIG. 11).

Then, referring to the conversion rule list,
Updates the certainty factor of the attribute (process result) list of the own object. In the acquired conversion rule list (16), "LI
GHT-RAIN is FAST, MODERATE-R
AIN is NORMAL, HEAVY-RAIN is SL
Convert to OW "is described. Therefore, the certainty factor 0 of ((FAST) 0) corresponding to ((LIGHT-RAIN) 0.9) in the list (23) stored in the attribute (processing result) of the own object is the previously acquired basic value. Updated to probability 0.9,
In addition, ((LIGHT-RAIN MODERATE-RAI
N) 0.1), which corresponds to ((FAST NORMAL) 0), is renewed with the certainty factor of 0 to the previously obtained basic probability of 0.1. By the above operation, the following list is obtained (step X4 in FIG. 11).

(((FAST) 0.9) ((FAST NORMAL) 0.1) ((NORMAL) 0) ((FAST NORMAL SLOW) 0) (29) ((NORMAL SLOW) 0) ((SLOW) 0) ((FAST SLOW) 0)) This processing result list (29) is stored in the attribute (processing result) of the own object (step X5 in FIG. 11).

Again, the process returns to the method selection, and the method (O33 in FIG. 3) of the next input (road congestion situation) object is selected according to the inference procedure (13) (step S2 in FIG. 5).

Then, the method 1 (step S in FIG. 5).
3) is completed (step S4 in FIG. 5), the method of the next (road congestion situation) object (O in FIG. 3).
43) is selected (step S2 in FIG. 5).

Similarly, method 2 (step S3 in FIG. 5)
Is executed (step S5 in FIG. 5), and the process returns to the selection of the method (step S2 in FIG. 5).

The input / translation process for the "road congestion" is the same as that for the "rainfall", and its details are omitted. However, the environmental value of "road congestion" is "time required to move 5 km away from the mesh point (unit:
It is assumed that “7 minutes” is input in this embodiment. FIG. 16 shows a situation where an environment value is input to the membership function of “road congestion situation”. The same processing as the basic probability allocation of the "rainfall" is performed, and the following focus element list (30) (((LIGHT- CROWDED) 0.9) ((LIGHT-CROWDED MODERATE-CROWDED) 0.1) ((MODERATE-CROWDED) 0) HEAVY-CROWDED) 0) ((LIGHT-CROWDED HEAVY-CROWDED) 0)) (30) is stored. The translation (road congestion) object (O4 in FIG. 3) is also subjected to the same processing as the translation (rainfall) object (O2 in FIG. 3), and its attribute (processing result) is Processing result list (2
9) is stored. Up to this point, the processing up to “4. Translation (road congestion status)” of the inference procedure (13) is completed,
The next integrated object method (O52 in FIG. 3) is selected (step S2 in FIG. 5).

Since the method type of the integrated object is the integrated type (step S3 in FIG. 5), method 3 is executed (step S6 in FIG. 5).

In the method 3 as well, first, the list (2
Initialization of the certainty factor in 4) is performed in the same manner as in methods 1 and 2 (step Y1 in FIG. 12).

Next, the certainty factor is acquired from the list of the attributes (process results) of the two upper objects. Here, FIG.
As can be seen from, it is important that only Integration Objects have two superordinate objects linked by direct links. That is, the certainty factor of the list (29) stored in the attributes (process results) of the translation (rainfall) object (O2 in FIG. 3) and the translation (road congestion state) object (O4 in FIG. 3) (( FAST) 0.9) ((FAST NORMAL) 0.1) is acquired (step Y2 in FIG. 12).

In the present embodiment, the certainty factor acquired from the translation (rainfall) object and the certainty factor from the translation (road congestion) object coincidentally.

Next, the two acquired certainty factors are integrated.
The state of this integration is shown in FIG. FIG. 17 shows that the products of the processing result list of the translation (rainfall) object and the processing result list of the translation (road congestion) object (29) whose focus element confidence is not 0 are multiplied. It has a shape. Then, the list (24) of the attributes (processing results) of the own object is updated as follows, and the processing result list is generated (step Y3 in FIG. 12).

(((FAST & FAST) 0.81) ((FAST & FAST FAST & NORMAL) 0.09) ... ((FAST & FAST NORMAL & FAST) 0.09)) , But the list for events with a confidence level of 0 is omitted.

Next, a method (O63 in FIG. 3) of the translation (integration result) object is selected (step S in FIG. 5).
2).

This method is a translation type (see FIG. 5).
Step S3), the same method 2 as the previous translation (rainfall) object and translation (road congestion situation) object
Is executed (step S5 in FIG. 5). The difference from the translation (rainfall) / translation (road congestion situation) object is that the list (18) is used as the conversion rule. According to this conversion rule list (18), (FAST & FAST) → (FAST), (FAST & FAST FAST & NORMAL) → (FAST NORMAL) ... → (FAST NORMAL NORMAL NORMAL) ... (*), so the consequent of (*) is the same (FAST NORMAL
AL), which means (FAST & FAST) → (FAST) (FAST & FAST FAST & NORMAL),
(FAST & FAST NORMAL & FAST), (F
AST & FAST NORMAL & FAST NORMA
L & FAST NORMAL & NORMAL) → (FA
(ST NORMAL). Therefore, by using the certainty factor of the list (31) of the attribute (processing result) of the higher-level object, the certainty factor of the list (23) of the attribute (processing result) of this translation (integration result) object is (((FAST) 0.81). ) ((FAST NORMAL) 0.19 (= 0.09 + 0.09 + 0.01)) ((NORMAL) 0) ((FAST NORMAL SLOW) 0) (32) ((NORMAL SLOW) 0) ((SLOW) 0) ((FAST SLOW) 0).

Next, the method of the interpretation object located at the end of the inference procedure (13) is selected (step S2 in FIG. 5). Since the method type of the interpretation object is the interpretation type (step S3 in FIG. 5), method 4 is executed (step S7 in FIG. 5). Also in method 4, first, the initialization of the upper and lower bound probabilities (confidence) of the list (25) stored in the attribute (process result) of the own object is performed as in the case of methods 1, 2, and 3. Perform (Fig. 13
Step Z1).

Next, the certainty factor is acquired from the list of the attributes (process results) of the upper objects. That is, ((FAST) 0.81) ((FAST NORMAL) 0.19) (33) is acquired from the list (32) of the attributes (process results) of the translation (integration result) object (step Z2 in FIG. 13).

Next, the upper and lower bound probabilities of each element of the universal set are calculated. Here, each element of the universal set is (FAS
T, NORMAL, SLOW). First, the upper bound probability of each element is calculated. FAST uses the list (3
Since it is included in both 3), the upper bound probability of FAST is 0.81 + 0.19 = 1.0 (34). Since NORMAL is included only in the lower list of the above list (33), the upper bound probability of NORMAL is 0.19. Since SLOW does not appear anywhere in the list (33) above, the upper bound probability of SLOW is zero. Next, the lower bound probability is calculated. FAST appears alone in the upper list above list (33), so F
The lower bound probability of AST is 0.81. NORMAL and SL
Since OW does not appear alone in either of the above lists (33), the lower bound probability of NORMAL and SLOW is zero. From the above, the list (25) of the attributes (process results) of the interpreted object updates the following list (35) (step Z3 in FIG. 13).

(((FAST) 0.81 1.0) ((NORMAL) 0 0.19) (35) ((SLOW) 0 0) As a result of this series of inference processes, the attributes (lists generated in each object are stored ( Figure 1 shows the contents of the processing result)
8 shows. In FIG. 18, an event with a certainty factor of 0 is omitted.

Up to this point, the case where the input value is given in the numerical form has been described. However, according to the present invention, not only the numerical format but also the fuzzy value as shown in FIG. 19 can be input. Now, assume that a fuzzy value such as the shaded portion in FIG. 19 is input (step B1 in FIG. 14).

This input value means that the amount of rainfall is 15 to 20 mm. When such a fuzzy input value is given, it is processed as follows.

In this case, the basic probability is calculated based on the maximum value of the membership function that intersects with the shaded portion of the input value. The membership functions intersecting with the shaded area in FIG. 19 are (LIGHT-RAIN) and (LIGHT
-RAIN MODERATE-RAIN). The maximum value of the points where each intersects is 1.0 for (LIGHT-RAIN), and (LIGHT-RAIN MODERATE-RAI).
N) is 0.29. Therefore, temporarily set these values to (LIGHT-
RAIN) and (LIGHT-RAIN MODERAT
E-RAIN) (step B in FIG. 14)
2).

Then, normalization is performed as in the case of FIG. That is, regarding (LIGHT-RAIN),

[Equation 10] Next, (LIGHT-RAIN MODERATE-
Regarding RAIN),

[Equation 11] become that way. Therefore, the focus element list (2
The basic probability of 1) is (((LIGHT-RAIN) 0.78) ((LIGHT-RAIN MODERATE-RAIN) 0.22) ((MODERATE-RAIN) 0) ((LIGHT-RAIN MODERATE-RAIN HEAVY-RAIN) 0) (( MODERATE-RAIN HEAVY-RAIN) 0) ((HEAVY-RAIN) 0) ((LIGHT-RAIN HEAVY-RAIN) 0)) (38) (FIG. 14, step B3).

In the above example, the translation objects O2 and O4 are installed behind the input objects O1 and O3 in FIG. 2 and the processing is performed in the order of input / translation / integration. , The translation objects O2 and O4 are deleted as shown in FIG.
By rewriting the conversion rule data stored in the attribute (conversion rule) 6 (O62 in FIG. 3) from the list (18) to the following list (39), the same processing as the above processing can be performed.

((LIGHT-RAIN-CROWDED FAST) (LIGHT-RAIN & MODERATE-CROWDED NORMAL) SLOW) (MODERATE-RAIN & HEAVY-CROWDED SLOW) (HEAVY-RAIN & MODERATE-CROWDED SLOW) (HEAVY-RAIN & HEAVY-CROWDED SLOW)) (39) Each object in the over diagrams, according to the preference of the problems and user handling, can be freely recombinant.

Next, another embodiment of the present invention will be described.

FIG. 21 is a diagram showing the structure of a fuzzy reasoning apparatus according to another embodiment.

The fuzzy reasoning apparatus shown in the figure differs from the fuzzy reasoning apparatus of the above-described embodiment in the following points. That is, in the fuzzy inference apparatus of the above-described embodiment, the inference procedure is preliminarily determined in the inference procedure generation unit 1 as a preprocessing before the execution of fuzzy inference, but in the fuzzy inference apparatus shown in FIG. The inference procedure is sequentially determined during execution of. Therefore, FIG.
In the ambiguous inference apparatus of this embodiment shown in FIG. 1, the inference procedure generation unit 1 is deleted from the ambiguous inference apparatus shown in FIG. 1, while the activation method searching unit 17 is newly added. Then, during execution of the fuzzy inference, the activation method search unit 17 receives a command from the inference execution management unit 3, and the object information 1
When 0 is searched, a method to be activated is found, and the method name is taught to the inference execution management unit 3, the inference execution management unit 3 activates the method. The fuzzy inference apparatus of this embodiment advances the fuzzy inference processing by repeating the above operation. This makes it possible to perform the inference process more quickly when, for example, the observation data is input intermittently instead of continuously. This is because the processing of the previously input observation data can be advanced before the next observation data is input.

The present invention is not limited to the above-mentioned embodiments, and can be carried out with various modifications without departing from the scope of the invention.

For example, "rainfall" and "road congestion"
Although an example of calculating the mobility cost based on these two factors has been shown, it is also possible to calculate the mobility cost by adding three or more observation data by adding "terrain inclination state" to the factors. It is possible.

Further, as in the “sales sales forecast problem”, the “regional population” and the “location of the store (distance from the main street or distance from the nearest station)” are integrated to calculate the forecast sales. The present invention is widely applicable to the problem of integrating a plurality of observation events and deriving a conclusion by using the DS theory such as calculation.

[0112]

As described above, according to the present invention, the device can be easily configured according to the problem to be handled and the type of observation data, and the addition of various procedures necessary for inference. Easy to modify knowledge. In addition, since the process during inference execution can be reduced to only the calculation of the certainty factor and the update process, inference can be speeded up. Furthermore, the processing of previously input observation data can be performed before the next observation data is input, so that, for example, when observation data is input intermittently instead of continuously. Inference processing can be performed more quickly.

[Brief description of drawings]

FIG. 1 is a block diagram showing the configuration of a fuzzy reasoning apparatus according to an embodiment of the present invention.

FIG. 2 is a diagram showing a data flow diagram in the embodiment.

FIG. 3 is a diagram showing details of a data flow diagram in the embodiment.

FIG. 4 is a diagram showing a membership function in the same embodiment.

FIG. 5 is a flowchart showing an operation procedure of inference processing in the embodiment.

FIG. 6 is a flowchart showing an initial setting operation procedure of inference processing in the embodiment.

FIG. 7 is a flowchart showing an initial setting operation procedure of inference processing in the embodiment.

FIG. 8 is a flowchart showing an initial setting operation procedure of inference processing in the embodiment.

FIG. 9 is a flowchart showing an initial setting operation procedure of inference processing in the embodiment.

FIG. 10 is a flowchart showing a method operation procedure of inference processing in the embodiment.

FIG. 11 is a flowchart showing a method operation procedure of inference processing in the embodiment.

FIG. 12 is a flowchart showing a method operation procedure of inference processing in the embodiment.

FIG. 13 is a flowchart showing a method operation procedure of inference processing in the embodiment.

FIG. 14 is a flowchart showing a fuzzy inference operation procedure of method 1 in the embodiment.

FIG. 15 is a diagram showing temporary assignment of the degree of belonging to each focus element in the same example.

FIG. 16 is a diagram showing temporary assignment of a degree of belonging to each focus element in the same example.

FIG. 17 is a diagram showing integration of two probability values in the same example.

FIG. 18 is a diagram showing attribute values of attributes (processing results) of each object in the embodiment.

FIG. 19 is a diagram showing the handling of fuzzy input values in the same embodiment.

FIG. 20 is a diagram showing a data flow diagram in which an object is changed in the embodiment.

FIG. 21 is a block diagram showing the configuration of a fuzzy reasoning device according to another embodiment of the present invention.

FIG. 22 is a diagram for explaining the basic concept of DS theory.

FIG. 23 is a diagram for explaining a Dempster connection rule.

FIG. 24 is an example of a membership function in fuzzy logic.

[Explanation of symbols]

 1 ... Inference procedure generation unit 2 ... Inference preprocessing execution unit 3 ... Inference execution management unit 4 ... Input method 5 ... Translation method 6 ... Integrated method 7 ... Interpretative method 8 ………… Fuzzy inference execution unit 9 ………… DS inference execution unit 10 ………… Object information 11 ………… Data flow diagram 12 ………… Universal set 13 ………… Conversion rule 14 ………… Membership function 15 ………… Environmental value 16 ………… Inference result

Claims (10)

[Claims] 1. A first data conversion means for converting observation data into a first data list with confidence, based on fuzzy inference; and a first data list with confidence, based on a conversion rule list. Second data conversion means for converting into a second data list with certainty factor, and a second data list with plural certainty factors corresponding to plural kinds of observation data based on Dempster's combining rule And a fourth data list with upper and lower bound probabilities from the third data list based on the Dempster-Shafer probability theory. A data calculation unit; a start order determination unit that determines a start order of the first data conversion unit, the second data conversion unit, the data integration unit, and the data calculation unit; And an execution management unit for sequentially activating and executing the first data conversion unit, the second data conversion unit, the data integration unit, or the data calculation unit according to a predetermined activation order. Fuzzy reasoning device. 2. A first data conversion means for converting observation data into a first data list with confidence, based on fuzzy inference; and a first data list with confidence, based on a conversion rule list. Second data conversion means for converting into a second data list with certainty factor, and a second data list with plural certainty factors corresponding to plural kinds of observation data based on Dempster's combining rule And a fourth data list with upper and lower bound probabilities from the third data list based on the Dempster-Shafer probability theory. A data calculation unit; a start order determination unit that determines a start order of the first data conversion unit, the second data conversion unit, the data integration unit, and the data calculation unit; A membership function necessary for fuzzy inference, an input means for inputting the model of the first data list with the certainty factor and the conversion rule list, and a first data conversion means according to the determined starting order A second data conversion means, a data integration means, a data calculation means, the input membership function,
Means for preliminarily associating the template of the first data list with certainty factor with the conversion rule list before executing fuzzy inference, and the associated activation order, membership function,
According to the template of the first data list with certainty factor and the conversion rule list, the first data conversion means and the second data conversion means.
An ambiguous inference apparatus, comprising: the data conversion means, the data integration means, or the execution management means for sequentially activating and executing the data calculation means. 3. A first data conversion means for converting observation data into a first data list with certainty factor based on fuzzy inference; and a first data list with certainty factor based on a conversion rule list. Second data conversion means for converting into a second data list with certainty factor, and a second data list with plural certainty factors corresponding to plural kinds of observation data based on Dempster's combining rule And a fourth data list with upper and lower bound probabilities from the third data list based on the Dempster-Shafer probability theory. Data calculation means and the first data conversion means during execution of fuzzy inference,
A deciding means for sequentially deciding a means to be activated and executed among the second data converting means, the data integrating means and the data calculating means, and an execution managing means for activating and executing the decided means. An ambiguous inference device comprising: 4. A first data conversion means for converting observation data into a first data list with certainty factor based on fuzzy inference; and a first data list with certainty factor based on a conversion rule list. Second data conversion means for converting into a second data list with certainty factor, and a second data list with plural certainty factors corresponding to plural kinds of observation data based on Dempster's combining rule And a fourth data list with upper and lower bound probabilities from the third data list based on the Dempster-Shafer probability theory. Data calculating means, input means for inputting the membership function necessary for the fuzzy inference, the template of the first data list with the certainty factor, and the conversion rule list, During execution of the inference, the first data conversion means,
Determining means for sequentially determining means to be activated and executed among the second data converting means, the data integrating means and the data calculating means, the determined means and the input membership function, An ambiguous inference device, comprising: an execution management unit that activates and executes the means while associating the template or the conversion rule list of the first data list with certainty factor. 5. A first data conversion means for converting observation data into a first data list with certainty factor based on fuzzy inference; and a first data list with certainty factor based on a conversion rule list. Second data conversion means for converting into a second data list with certainty factor, and a second data list with plural certainty factors corresponding to plural kinds of observation data based on Dempster's combining rule And a fourth data list with upper and lower bound probabilities from the third data list based on the Dempster-Shafer probability theory. A method of appropriately combining the data calculation means with the first data conversion means, the second data conversion means, the data integration means or the data calculation means according to the content to be inferred. Fuzzy inference apparatus characterized by comprising and. 6. A first data conversion means for converting observation data into a first data list with certainty factor based on fuzzy inference, and the first data list with certainty factor based on a conversion rule list. Second data conversion means for converting into a second data list with certainty factor, and a second data list with plural certainty factors corresponding to plural kinds of observation data based on Dempster's combining rule And a fourth data list with upper and lower bound probabilities from the third data list based on the Dempster-Shafer probability theory. The data calculation means and the entire inference process that integrates multiple types of observation data are divided into modules, and a data flow diagram that connects each module with a relational link is input. And a means for appropriately combining the first data conversion means, the second data conversion means, the data integration means or the data calculation means according to the input data flow diagram. Ambiguous reasoning device. 7. The fuzzy inference apparatus according to claim 1 or 6, wherein at least a plurality of observation data composed of meteorological data and map data of a predetermined area and a corresponding area based on knowledge prepared by a human in advance. A fuzzy reasoning device characterized in that the evaluation value of is comprehensively calculated as the result of regional analysis. 8. A first procedure for converting observation data into a first data list with certainty factor based on fuzzy inference, and the first data list with certainty factor for certainty factor based on a conversion rule list. Based on the second procedure of converting to the second data list with the certainty and the Dempster's combining rule, the second data list with the certainty factor corresponding to the multiple types of observation data is added with the certainty factor Third procedure to integrate with the third data list of Dempster-Shafer
The fourth procedure for calculating the fourth data list with the upper bound probability and the lower bound probability from the third data list based on the probability theory of is prepared as a component and prepared in advance according to the content to be inferred. An ambiguous inference method characterized by appropriately combining these first to fourth procedures. 9. A first procedure for converting observation data into a first data list with certainty factor based on fuzzy inference, and the first data list with certainty factor for the certainty factor based on a conversion rule list. Based on the second procedure of converting to the second data list with the certainty and the Dempster's combining rule, the second data list with the certainty factor corresponding to the multiple types of observation data is added with the certainty factor Third procedure to integrate with the third data list of Dempster-Shafer
The fourth procedure for calculating the fourth data list with the upper bound probability and the lower bound probability from the third data list based on the probability theory of is prepared as a component and prepared in advance according to the content to be inferred. An ambiguous inference method is characterized in that the starting order of these first to fourth procedures is determined, and the first to fourth procedures are sequentially started to perform inference in accordance with the determined starting order. 10. A first procedure for converting observation data into a first data list with certainty factor based on fuzzy inference, and a first procedure with certainty factor based on a conversion rule list.
Based on the second procedure of converting the data list of to the second data list with certainty factor and Dempster's combining rule,
A third procedure for integrating a plurality of types of the second data list with confidence corresponding to a plurality of types of observation data into a third data list with confidence, and Dempster-Shafe
A fourth procedure for calculating a fourth data list with upper and lower bound probabilities from the third data list based on the probability theory of r is prepared as a component and prepared in advance, and ambiguous inference is being performed. In addition, the ambiguous inference method is characterized in that among the first to fourth procedures, the procedure to be activated and executed is sequentially determined, and the determined procedure is activated and executed.