JPH0458331A - Interface device between approximate inference device and human being - Google Patents

Abstract

PURPOSE:To obtain a result accordant with the characteristic of a machine by multiplying the possibility value of the conclusion obtained from an approximate inference device by the set weight of conclusion to calculate the score of each conclusion and correcting the weight when no coincidence is obtained between the order of calculated scores and the priority set by a human being. CONSTITUTION:A means 35 is provided to set the weight to each conclusion together with a means 32 which performs a prescribed arithmetic operation between the possibility of a conclusion and the weight set for the conclusion and calculates the score of each conclusion, a means 34 which inputs the priority when the coincidence is obtained between the order of scores and the priority set by a human being, and a means 37 which changes the weight in order to secure the coincidence between the order of scores and the input priority. Then the weight is applied to the possibility value of the conclusion obtained from the approximate inference device so that the score is calculated for each conclusion. Then the weight is changed so as to secure the coincidence between the order of scores and the priority set by the human being. Thus it is possible to obtain a conclusion that is accordant with the characteristic of an inference subject or the convenience of the human being.

Description

[Detailed Description of the Invention] Setting the weight of the summary conclusion of the invention. A score is calculated for each conclusion by multiplying the probability value of the conclusion obtained from the approximate inference device by the weight set for that conclusion. When the order of the obtained scores is different from the priority order thought by humans, the weights are corrected so that the order of the scores matches the priority order.

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an approximate inference device and a human interface device.

Prior Art Approximate inference is known in which the inference result is modified or changed depending on the amount of information about the event used to derive the inference result (for example, Zhangl (onHt n
AN EXPERT SYSTEM WITHTH
INKING INIMAGEs”, Preprin
ts of 5econd IFSA Congress
s.

Tokyo, July 20-25.1987. P
, 765).

This approximate inference method calculates the amount of information for each event (i.e., information identification ability of the event) using a membership function given to each conclusion for the event, and calculates the inference result (i.e., the probability that the conclusion holds true). The aim is to improve the discriminative ability of inference results by modifying or changing the information based on the amount of information of the event used to draw the conclusion (by multiplying the probability and the amount of information).

However, in such approximate reasoning devices, the weight of the conclusion itself (for example, in the case of fault diagnosis, the cost and time required for repair, etc.) is not taken into consideration.
This can be detected in cases where it is difficult to know which conclusion is important when two conclusions with almost the same probability are reached, and there is a conclusion that should be given priority even if the probability is not high. The problem is that there is no. Therefore, it is not possible to make inferences that match the characteristics of the inference target, and in order to match approximate inference to the characteristics of the inference target system (equipment),
It is necessary to change the knowledge base of the approximate inference device itself.

SUMMARY OF THE INVENTION OBJECT OF THE INVENTION The present invention provides an interface device that can adapt the conclusion obtained from an approximate inference device to the characteristics of the inference object, the convenience of the person, etc.

Structure of the Invention 1 Operation and Effect The approximate inference device and human interface device according to the present invention calculates the probability of a conclusion according to given input data using knowledge representing the relationship between a plurality of events and a conclusion. Provided for approximate reasoning devices. And the interface device.

A means for setting a weight for each conclusion, a means for calculating a score for each conclusion by performing a predetermined operation between the possibility of a conclusion and the weight set for that conclusion, and a means for calculating a score for each conclusion. The present invention is characterized in that it includes means for inputting the priority when different from the priority, and means for changing the weight so that the order of the magnitude of scores matches the input priority.

There is also an interface device according to the invention.

A means for setting a weight for each conclusion, a means for calculating a score for each conclusion by performing a predetermined operation between the possibility of a conclusion and the weight set for that conclusion, and a means for calculating a score for each conclusion, The present invention is characterized in that it includes an input means for modifying the weights so that they match when the priorities are different from the considered priorities.

According to this invention, a score is calculated for each conclusion by applying weight to the probability of the conclusion obtained from the approximate inference device. Then, the weights are changed so that the order of the scores matches the priorities considered by humans.

Therefore, the knowledge base of the approximate inference device does not need to be changed to match the characteristics of the inference target. Or, it is possible to reach a conclusion that suits human convenience.

Description of Embodiments (1) Overall configuration of approximate inference device and interface device FIG. 1 shows an example of the overall configuration of the approximate inference device and interface device. The approximate inference device includes a knowledge storage device 11. Knowledge synthesis device 121 Post-synthesis knowledge storage device 1
3. Event value input device 14. Fitness calculation device 15. Dynamic information amount calculation device 1B, possibility calculation device 17. Static information amount calculation device 19. Clarity calculation device 20, clarity storage device 21
．． It consists of a clarity adding device 22 and a clarity display device 23.

The interface device between the approximate inference device and humans (workers, staff, etc.) is one weight storage device 31. Score calculation device 32.
Score display device 33. It is composed of a correction input device 341, a weight change amount setting device 351, a weight change amount storage device 36, and a weight change device 37.

Below, these devices will be explained in detail using failure diagnosis as an example.

(2) Knowledge storage device The knowledge storage device 11 stores knowledge input by experts in a format that shows the relationship between events (situations caused by failures, measurement results, etc.) and conclusions (types of failures, etc.). It is a device. This device can store the knowledge of multiple experts.

Two experts ext stored in the knowledge storage device 11
, ex2 knowledge example is shown below in the form of a rule.

Expert exl: if 20≦f1≦60, 0≦r2≦40. the
n cl-=(1) if40≦fl≦80.80≦f
2≦100. then c2-(2) fl and f2 are events, and these are sometimes called event 1 and event 2, respectively. cl and c2 are conclusions, and these are sometimes called conclusions 1 and 2, respectively□.

Further, a and b expressed as a≦r1≦b are called a minimum value and a maximum value, respectively.

The above rules are expressed in the form of a table for each expert as follows.

Expert ex2: if 30≦f1≦50. 10≦f2≦30.
then cl-(3) if 50≦f1≦70.
70≦f2≦90. then e2・ (4)
(3) Knowledge synthesis device The knowledge synthesis device 12 is a device that synthesizes the knowledge of a plurality of experts stored in the knowledge storage device 11 and combines it into one piece of knowledge.

There are various ways to synthesize knowledge, but here we calculate the average value and standard deviation of multiple experts for the maximum and minimum values of each event involved in each conclusion.

The knowledge synthesis process will be explained below, taking as an example the knowledge of the two experts described above that leads to the conclusion CI from the event f1.

From the above rules (Equations (1) and (3)), event 1
The rule for obtaining conclusion 1 (cl) from (fl) is extracted as follows.

Expert ext: if 20≦fl≦fiOthe
n cl-(5) Expert ex2: if 30≦
f1≦50 then cl-(8) The average value “if□” of the minimum value and the average value m of the maximum value are calculated.

■ax Standard deviation of the minimum value σ and Standard of the maximum value in This type of composite operation of expert knowledge is performed using the rules described above (
For equations (1) to (4), if all the minimum and maximum values of each event involved in each conclusion are examined, a table as shown below is obtained.

Table 3 Generally, in approximate reasoning, a membership function is given to an event. Here, as an example, a method for determining membership functions using Gaussian distribution using expert knowledge synthesized as described above will be described.

Average value m of minimum values, average value m of maximum values.

Using the standard deviation σ of the minimum value, and the standard deviation Ain σ of the maximum value, the membership function is calculated by the following formula: l1a
x

Only the left half is used. Φ(x)-X at 0,5
The position of is determined by m or mff1l
n 1laX, the slope is σ,
or determined by σff1ln
IaXru.

As an example, the membership function for determining the conclusion c1 from the event fl is created as shown in FIGS. 3a to 3c using the values calculated by equations (7) to (10). In this case, equation (11) becomes as follows.

...(11) however.

X: Value of input data to the event Φ(X) Degree to which human input data X fits the event (degree of fit) Gauss(x): Value of Gaussian distribution in input X.

Figure 2 shows an example of a Gaussian distribution. In order to create the membership function in this Gaussian distribution, Figure 3a shows (
Figure 3b shows the first term on the right side of equation (11) or equation (12), Figure 3c shows the second term on the right side of equation (11) or equation (12), and Figure 3c shows the first term to the second term above. represents the result of subtracting , that is, the membership function represented by equation (11) or equation (12).

Examples of membership functions for determining conclusions cl and c2 for each event fl and f2 created based on the synthesized knowledge shown in Table 3 are shown in Figures 4a and 4b.
As shown in the figure.

(4) Post-synthesis knowledge storage device The post-synthesis knowledge storage device 13 stores the average value and standard deviation calculated by the knowledge synthesis device 12 in the format shown in Table 3. Since it is not necessary to synthesize knowledge every time an inference is made, the results of preliminary calculations like this are memorized. By reading out and using the values in the storage device 13 each time an inference is made, it is possible to speed up the inference process.

(5) Event value input device The event value input device 14 is a failure diagnosis target device, a keyboard 9 communication interface device, and a memory.

This is a device that reads input data input for each event from a file or the like. The input data is given to the fitness computing device 15, and information as to whether data for each event has been input is given to the clarity addition device 22.

(6) Fitness calculation device The fitness calculation device 15 calculates the fitness of the data input from the event value input device 14 for each membership function (or conclusion). Specifically, the goodness of fit is obtained as Φ(X) by substituting the input data as the variable X on the right side of equation (11). Of course, such an arithmetic expression does not necessarily have to be used.

(7) Dynamic information calculation device and static information calculation device event f
The event value (input data) of event 1 is xl, and the event value of event f2 is x2. These data are input from the event value input device 14.

As shown in Figures 5a and 5b, each degree of fitness is "11".
"12, m21' m22 is defined as follows.

m1□: goodness of fit m1 of input data x1 to conclusion c1
2: Degree of fit of input data x1 to conclusion c2 m2□:
Degree of suitability of input data x2 to conclusion c1 m22: Degree of suitability of input data x2 to conclusion c2 These degrees of suitability are calculated by the suitability calculation unit 15 when input data xl and x2 are given.

Let us now consider the concept of fuzzy entropy.

Fuzzy entropy Ef when input x1 is given
1 is defined as follows.

This fuzzy entropy is a type of index of information discrimination ability, and when input data x1 is given, the value is small enough that the conclusion can be clearly identified, and the value is large enough that the conclusion can only be vaguely identified. In other words, input data x
The larger the difference between the goodness of fit "11" for the conclusion C1 of input data x1 and the goodness of fit m12 for the conclusion C2 of input data x1, the smaller the value becomes, and the smaller the difference, the larger the value becomes.

Similarly, when the input x2 is given, the fuzzy entropy Er2 is given by the following equation, and the range of values that the fuzzy entropy Ef can take is as shown below.

0 ≦ Ef ≦ log<n ) n, the number of conclusions on the event. In this example, the number of conclusions on the event 1 (rl) is 2 (cl.

c2), the maximum value of the fuzzy entropy Ef is log(2).

Next, using this fuzzy entropy Efl, the amount of dynamic information 1flD (
Find Xi). Here, the amount of dynamic information 1flo(xi)
is the ability to identify events to determine a conclusion when making inferences, and the larger the difference between the degree of conformity ``11'' of input data x1 to conclusion c1 and the degree of conformity m12 of input data x1 to conclusion c2, the larger the value The smaller the difference, the smaller the value.

Therefore, the amount of dynamic information 1flD(xi
) from the maximum fuzzy entropy, the input data xi
is defined as minus the fuzzy entropy Efl when is given.

(Hereafter, margin) The amount of information is given by the following equation.

Ifslff5-1O Similarly, the amount of dynamic information when input data x2 is given for two events r2 is defined as follows.

The dynamic information amount calculating device 1B uses the fitness obtained by the fitness calculating device 15 to calculate the dynamic information amount for each event according to equations (15) and (16).

The amount of dynamic information depends on the input data xi, x2 as described above. On the other hand, the amount of static information does not depend on the input data, and from the maximum fuzzy entropy,
The static information amount of the entire event is determined by subtracting the average fuzzy entropy within the range of the event. for example,
Static information about event 1... (17) Similarly, the amount of static information about event 2 is calculated using the following formula % Formula % (): (): Fit of input data X to conclusion cl for event f1 Regarding event fl The goodness of fit of the input data X to the conclusion c2 m2t(x): The goodness of fit of the input data Calculate the fuzzy entropy for each X at different intervals, and calculate the average of them (0 and δ≦100) As can be seen from equations (17) and (18), the event The larger the overlap between the membership functions of the event, the smaller the static information amount of the event, and the smaller the overlap between the membership functions of the event, the larger the static information amount of the event. In other words,
The amount of static information indicates the ability of an event's membership function to identify a conclusion.

The static information amount calculating device 19 calculates and stores the amount of static information for each event from the membership function obtained from the synthesized knowledge according to the above-mentioned equations (17> and (18)). Since the static information amount does not depend on input data, it only needs to be calculated once.

(8) Possibility calculation device For each conclusion, calculates the amount of information of events such that the sum of the amount of information of events involved in that conclusion is 1, and the relative strength of the amount of information of those events does not change. . This calculated amount of information is called a weight.

For example, if the above-mentioned dynamic information amount is used, each weight will be as follows.

Weight of event 1 relative to conclusion 1: 'It ft , -I
fI (xi')/E 1flD(xi)+I
f2D(x2) co-(19) Weight of event 2 relative to conclusion 1: Wf12Ir2(x2)/[0D(xi)+
1f2D(x2)] - (20) Weight of event 1 for conclusion 2: Wf21-IN (xi)/[IflD(
xi) + If2D (x2)] - (21) Weight of event 2 for conclusion 2: wf2° - If'2 (X2)/
[1No(xi)+1f2D(x2>]...(
22) Next, calculate the product of these weights and the goodness of fit.

The sum of these results for each conclusion is calculated as the probability of the conclusion.

For example, in the above example, possibility of conclusion 1 = wf Xm + wf Xm
- (23) Possibility of conclusion 2 - wf xm +
wf xm - (24).

The possibility calculation device 17 performs the above-mentioned calculation to calculate the probability for each conclusion.

(9) Weight Storage Device The weight storage device 31 stores the weights set for each conclusion in the form of a table as shown in Table 4 in a memory, file, or the like. Also.

The weight changed by the weight changing device 37 is captured and stored. The initial value of the weight of each conclusion is 1.0.

The weight of the conclusion may be specified by a human.

Table 4 (10) Score calculation device The score calculation device 32 calculates between the probability value of each conclusion calculated by the possibility calculation bag w17 and the weight of each conclusion stored in the weight storage device 3I. For example, the following equation is calculated to find the score for each conclusion.

p, -v, xw,
...(25)+11 Here, P is the score of conclusion cj (i-1, 2゜3, ...
・) ■ is the possibility of the conclusion ci (Equations (23) and (24)); W is the weight of the conclusion c1 (see Table 4) ■.

By this operation You can get it every time.

The scores as shown in Table 5 are the conclusions Table 5 (11) Score display device The score display device 33 displays the scores calculated by the score calculation device 32 for each conclusion to inform the human being. This device 33 may be one that transmits the scores through communication and stores the scores in a memory, file, or the like. In this embodiment, the top three conclusions with the highest scores are displayed for each conclusion. The number of conclusions to be displayed may be one or more than three.

(12) Correction input device The correction input device 34 is used to input the 3 points displayed on the score display device 33.
and the order of magnitude of the scores for each conclusion (fault).

This is used when the priority of actually repairing the failure (or the importance of the failure, etc.) is different, and is used to input what one person considers to be the correct order. If the size order of scores matches the priority considered by humans, there is no need to input it.

Here, what is the priority for actually repairing the failure?

It is determined by taking into account the number of man-hours and costs involved in repairing, the reliability of machines and parts, etc. For example, prioritize things that require as little man-hours as possible.

Previously, only the possibility of a conclusion was displayed, and the order of repair of failures was determined by considering the priority of one person. According to this invention, by inputting the priorities in advance by a human, it is possible to automate even decisions that take priorities into consideration. Furthermore, since the priority is entered by a human, it is possible to obtain results that match the characteristics of the machine at that time, and it is also possible to respond to changes in the machine over time.

(13) Weight change amount setting device This weight change amount setting device 35 sets (inputs) the amount of weight change (ratio) used for weight change in the 8 weight change device 37 in the form of a table as shown in Table 6. It is for. In this embodiment, as shown in Table 6, the amount of change for increase and decrease is determined depending on the magnitude of the conclusion possibility value. The amount of change in this weight can be changed by one person at any time.
Can be set and changed.

Assume that it is the preface of Table 6. The size of the score is cl, c2. c.
Since the order is 3, it is necessary to change the weights.

Table 7 (14) Weight change amount storage device This weight change amount storage device 3B is 1 weight change amount setting device 3
The amount of change in weight set in step 5 is stored in a table format such as Table 6. The change amount stored in this device 3G is referred to when changing the weight using the 1 weight change amount W37.

(15) Weight change device This weight change device 37 corrects and changes the weight of the conclusion according to the correct order (priority) corrected by a human input from the correction input device 34.

For example, the order (correction order, priority) input by one person is c2. as shown in Table 7. The forward weights of cl and a8 are changed according to the following algorithm.

1) Change the order of modification to increase the weight of the conclusion whose number is first. Among the increase change amounts stored in the weight change amount storage device 36, the increase change amount corresponding to the probability value of the conclusion is referred to. In this example, the probability of conclusion c2 is 0.7, so from Table 6, the amount of weight change is 1.1. Therefore, the current weight 1.
The product of 1 and the change amount 1,1 is set as a new weight wo.

c2: WO2-1, 1×1. ．． 1-1.21
...(2B) 2) If the modification order is 2nd, the weight of the conclusion is not changed. That is, cl: WOI −1,L
...(27) remains the same.

3) Change the modification order to reduce the weight of the conclusion number 3. Among the decrease change amounts stored in the weight change amount storage device 36, the decrease change amount corresponding to the probability value of the conclusion is referred to. In this example, the probability of C3 is 0.6, so from Table 6, the amount of weight change is 0.9. Therefore, the current weight 1,
Let the product of 0 and the change amount 0.9 be the new weight WO.

] c3: WOs 1.0 ×0.9-0.9
(28) The new weight WO9 calculated and changed in this way is stored in the weight storage device 31 through a communication interface device memory file or the like or directly.

In the case of the above example, the new weights and scores are as shown in Table 8, and the order of magnitude of the new scores matches the priority (correction order) input by the human.

Table 8 If the magnitude of the score does not match the priority input by the human, the above algorithm may be repeated again.

It goes without saying that the algorithm for changing the weight of the conclusion is not limited to the above example, and there are various methods.

As described above, the interface device allows inference results that take into account the characteristics of the five machines and match human thinking to be obtained without modifying the knowledge base itself.

In the above example, the weight of each conclusion is changed by one person inputting the order designation when the order of the displayed score size is different from the priority that humans think, but two people change the weight of each conclusion. It is also possible to directly change the weight itself. In this case.

It is advisable to provide a weight input device that takes into account the ease of manual input. This allows five people to modify the weights as appropriate. For example, if a staff member believes that by repairing a machine, it is less likely to cause a malfunction, it is possible to reduce the weight of the malfunction and adapt it to the characteristics of the machine. Become.

(16) Clarity calculation device The clarity calculation device 20 is a device that calculates the clarity of each event for each conclusion. Here, the clarity of each event for each conclusion indicates the relative discernment ability of each event when determining a certain conclusion.

Therefore, this clarity makes it possible to compare the discriminative ability of multiple events in order to establish a certain conclusion, and to understand which event has a high discriminative ability (contains a large amount of information). Ru. The method for calculating clarity is described below.

First, Table 9 shows the relationship between conclusions, events, and the amount of static information.

Table 9 As can be seen from Table 9, it also depends on the amount of static information.

It is possible to compare the discriminative ability of multiple events to establish each conclusion. However, as it is difficult to intuitively understand the relative discrimination ability as it is, we normalize the amount of static information for each conclusion as shown in the table below, and use the normalized value to clearly distinguish each event for each conclusion. CΩ.

Table 10 however.

CI −CD −1fl / (lfl + If
2s) 11 12 S 5 Cj7 - CI -IF5 / (1fls+ 1
f28) 21 22 S.

In this way, the clarity calculation device 20 calculates the clarity of each event for each conclusion.

(17) Clarity Storage Device The clarity storage device 21 is a device that stores the clarity of each event for each conclusion calculated by the clarity calculation device 20.

Clarity calculations do not need to be performed every time an inference is made. Therefore, the clarity calculated when knowledge is synthesized is stored in the clarity storage device 21, and the value stored in the clarity storage device 21 is read out every time an inference is made. This makes it possible to speed up the inference processing.

(18) Clarity addition device The clarity addition device 22 is a device that calculates the clarity of an event to which data is actually input. Here, for the inference that is actually made, the summation of the clarity of the events for which data is input is taken. The sum of this clarity indicates the clarity of the inference result. It can be said that the higher the clarity, the greater the amount of information for deriving the inference result. Therefore, clarity can be used as an indicator to judge the reliability of the inference result itself.

The clarity of the inference result is calculated as follows.

a) When data is input only for event 1 (fl) - Clarity for the inference result of conclusion 1 (cl) C, O,
−”11 ・Clarity cn2 for the inference result of conclusion 2 (c2)
C''12 b) When data is input only for event 2 (f2) - Clarity for the inference result of conclusion 1 (cl) Cρ1-
CΩ21 ・Clarity Cg2- for the inference result of inference 2 (c2)
”22 C) When data is entered for both Event 1 (fl) and Event 2 (r2) - Clarity for the inference result of Conclusion 1 (cl) C111
-CI +Cjl121-1.0 ・Clarity CΩ2- for the inference result of conclusion 2 (c2)
CI + cg22-i,.

What is the possible range of clarity Cg of the inference result?

00≦CI! ≦1.0. In other words, if you make an inference by inputting data about all the events that can be used to draw a certain conclusion within the knowledge given before making the inference,
The clarity of the conclusion will be 1.0. Also, only some of the events that can be used to draw a certain conclusion.

If the data is entered, clarity will be a value between 0.0 and 1.0. At this time, it can be said that if a large number of events with high clarity are used among the events that can be used, the clarity of the conclusion will also be increased, and a highly reliable inference result can be obtained.

(19) Clarity Display Device The clarity display device 23 is a device that displays the clarity of the inference result (-the possibility mentioned above as an example) calculated by the clarity addition device 22. The clarity may be displayed along with the inference results, or the clarity may be transmitted to another device or stored in memory or a file.

This clarity display displays all conclusions of the inference results. Therefore, if there are multiple conclusions, the clarity corresponding to each conclusion is displayed.

In this way, every time data is entered.

By calculating the information amount of the event to which the input data belongs and displaying the clarity of the inference result, the user can judge the reliability of the inference result.

It goes without saying that each of the devices 11 to 23 and 31 to 37 described above can be realized by a computer including a memory and a display device. For example, knowledge synthesis device 12° various calculation devices 15.16.17.19.20. ．． 22.32.37
is suitably realized by a CPU that operates according to a program.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an example of the overall configuration of an approximate inference device and an interface device. FIG. 2 is a graph showing a Gaussian distribution. Figures 3a to 3c are graphs showing how membership functions are formed. Figures 4a and 4b are graphs showing the membership functions obtained for each event. FIGS. 5a and 5b are graphs showing how the goodness of fit is determined. 11...Knowledge storage device. 12...Knowledge synthesis device. 13... Post-synthesis intelligence storage device. 14...Event value input device. 15...Fitness calculation device. 1B...Dynamic information calculation device. 17... Possibility calculation device 31... Weight storage device 32... Score calculation device. 33...Score display device. 34... Correction input device. 35... Weight change amount setting device. 3B... Weight change amount storage device 37... Weight change device.

Claims (2)

[Claims] (1) Using knowledge that represents the relationship between multiple events and conclusions,
A means for setting a weight for each conclusion is provided for an approximate inference device that calculates the probability of a conclusion according to given input data, and a means for setting a weight for each conclusion, and a means for setting a predetermined value between the probability of a conclusion and the weight set for that conclusion. A means for calculating the score of each conclusion by performing the calculation, a means for inputting the priority when the order of the magnitude of the score is different from the priority considered by humans, and a means for inputting the order of the magnitude of the score. An interface device between an approximate inference device and a human, comprising means for changing weights to match priorities. (2) Using knowledge that expresses the relationship between multiple events and conclusions,
A means for setting a weight for each conclusion is provided for an approximate inference device that calculates the probability of a conclusion according to given input data, and a means for setting a weight for each conclusion, and a means for setting a predetermined value between the probability of a conclusion and the weight set for that conclusion. means for calculating the score of each conclusion by performing the calculation of
and an input means for modifying the weights so as to make them match when the order of the magnitude of the scores is different from the priorities considered by humans.