JPH0546428A - Fuzzy diagnostic processor - Google Patents

Abstract

PURPOSE:To easily specify a failure cause by fuzzy operation of a fault phenomenon and membership functions. CONSTITUTION:Strengths of causal relations between fault causes and fault phenomena are held as membership functions in a membership function storage means 110, and a fault data gathering means 111 recognizes a fault phenomenon to collect it as fault data expressed by a numerical value, and a processor 112 performs the preliminarily determined fuzzy operation of fault data and membership functions to output analysis data where strengths of causal relations of plural fault causes to fault data are expressed by numerical values, and a display means 113 takes analysis data as the input and displays the strength of causal relations to each fault causes based on analysis data.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a diagnostic processor used in a computer system including a plurality of devices such as a central processing unit and peripheral control unit, and more particularly to a fuzzy diagnostic processor.

[0002]

2. Description of the Related Art Conventionally, this type of diagnostic processor does not have a membership function indicating the strength of the causal relationship between a failure phenomenon and a failure cause, and a plurality of failure causes are expected for the failure phenomenon. In this case, assuming that all of them have the same causal relationship, either remove all the probable cause of failure or leave it to the experience of the maintenance personnel to select and eliminate the cause of failure. The method was taken.

[0003]

In the above-described conventional diagnostic processor, the failure phenomenon that occurs as the computer system becomes more complicated, and it becomes very difficult to identify the cause of the failure. There are drawbacks in that all the components of the system are eliminated, and the cause of the failure cannot be identified due to the lack of experience of maintenance personnel, and it takes a long time to recover from the failure.

In view of the above points, an object of the present invention is to predefine a membership function indicating the strength of a causal relationship between a failure cause and a failure phenomenon, and to perform a fuzzy operation between the failure phenomenon and the membership function. It is to provide a fuzzy diagnostic processor that can easily identify the cause of a failure.

[0005]

A fuzzy diagnostic processor of the present invention is a diagnostic processor used in a computer system composed of a plurality of devices such as a central processing unit and a peripheral control unit, and may occur in the computer system. Holds a membership function in which the strength of the causal relationship between each of the multiple expected causes of failure and the multiple failure phenomena that are expected to be observed simultaneously or independently due to these causes of failure is numerically defined in advance. Membership function storage means, failure data collection means for recognizing a failure phenomenon occurring in the computer system and collecting it as failure data represented by a numerical value, and inputting the failure data and the membership function into fuzzy theory A predetermined fuzzy based on the failure data and the membership function based on An arithmetic processing unit that performs arithmetic processing and outputs analysis data that numerically expresses the causal relationship strength of each of the plurality of failure causes with respect to the failure data as a result of the operation processing; Display means for displaying the strength of a causal relationship based on the analysis data.

[0006]

In the fuzzy diagnostic processor of the present invention, it is expected that the membership function storage means will be observed at the same time or independently due to a plurality of failure causes expected to occur in the computer system and these failure causes. A membership function that defines in advance the numerical value of the causal relationship with each of a plurality of failure phenomena is held, and the failure data collection means recognizes the failure phenomenon that has occurred in the computer system and collects it as failure data represented by a numerical value. Then, the arithmetic processing unit inputs the failure data and the membership function, performs predetermined fuzzy operation processing on the failure data and the membership function based on the fuzzy theory, and performs a plurality of failure data on the failure data as a result of the operation processing. Outputs analysis data that numerically expresses the causal relationship strength of each cause of failure, and the display means displays the analysis data. Input to display the strength of the causal relationship based on the analysis data by failure cause.

[0007]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Next, the present invention will be described in detail with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of a fuzzy diagnostic processor 104 according to an embodiment of the present invention.
The fuzzy diagnostic processor 104 of this embodiment is shown applied to the diagnosis of a fault detected in the communication control unit (CCU) 102 in the computer system. In FIG. 1, reference numeral 101 is a central processing unit (CPU) that controls the entire computer system, performs arithmetic processing, etc., 102 is a communication control unit (CCU) having a normal communication control function, and 103 is a partner station. Indicates. CPU10
1 transfers a series of character data called a text with the partner station 103 via the CCU 102.

The fuzzy diagnostic processor 104 has a plurality of failure causes that are expected to occur in the computer system and a plurality of failure phenomena that are expected to be simultaneously or independently observed due to these failure causes. Membership function storing means 110 for storing a membership function in which the strength of causality is defined in a numerical value in a predefined format, and a failure phenomenon occurring in a computer system is recognized and collected as failure data represented in a numerical value. The fault data collection means 111 for inputting the fault data and the membership function are input, and predetermined fuzzy arithmetic processing is performed on the fault data and the membership function based on the fuzzy theory, and the fault data is obtained as a result of the arithmetic processing. Analysis data that numerically expresses the strength of the causal relationship for each of the multiple causes of failure. Display processor 112 as an arithmetic processing unit for outputting, the strength of the causal relationship based on the analysis data by failure cause enter the analysis data display means 1
13 and 13.

Reference numeral a is a higher-level interface that electrically and logically connects the CPU 101 and the CCU 102, and realizes transfer of commands and data between the CPU 101 and the CCU 102.

Reference numeral b denotes a communication line, which is connected to the partner station 103 and C.
An interface that electrically and logically connects to the CU 102, such as RS232C.

Reference numeral c is a diagnostic interface for sending a failure phenomenon occurring in the CCU 102 to the fuzzy diagnostic processor 104. Although not shown, the diagnostic interface c also exists for the components of the computer system, such as the CPU 101. The diagnostic interface c is connected to the fault data collecting means 111 in the fuzzy diagnostic processor 104 and collects the phenomenon data in the CCU 10.
2 to the fault data collection means 111. here,
The phenomenon data is information such as various error detection signals detected by the CCU 102.

D is a fault data interface,
The fault data collection unit 111 transfers the fault data digitized into a predefined format.

E is a function interface, which transfers the membership function to the processor 112.

Reference numeral f denotes a display interface, which transfers the analysis data, which is the result calculated by the processor 112, to the display means 113.

Next, a fuzzy operation which is a characteristic operation of the fuzzy diagnosis processor 104 of this embodiment will be described.

The total set X of failure causes that can be expected to occur is from m (positive integer) elements x _{1 to} x _m , and the total set Y of observable failure phenomena is n (positive integer) elements y. _1- y
_If it consists of _n , it can be described as X = {x ₁ , x ₂ , ..., X _m }, Y = {y ₁ , y ₂ , ..., Y _n }.

It is also clear that there is a causal relationship between each element x _i of the set X and each element y _j of the set Y.
Here, each causal relationship between the element x _i and the element y _i is represented by R _ij
If the strength of the relationship is represented by a real number from 0 to 1, R can be represented by a matrix of m rows and n columns, and this will be referred to as a membership function.

Now, if a failure occurs for some reason and a failure phenomenon is observed as a result, a fuzzy set X for the failure cause and a fuzzy set Y for the failure phenomenon.
It is known that each relationship can be described as Y = X * R by using the fuzzy operation represented by the membership functions R and *.

Therefore, it is also a known fact that the set X of failure causes can be specified by specifying the membership function R in advance and observing the set Y of failure phenomena.

Next, the operation of the fuzzy diagnostic processor 104 from the failure phenomenon to the specification of the failure cause will be specifically described.

First, as a precondition, a set X of failure causes is set.
X _m and the element y _n of the set Y of failure phenomena, m = 2 and n = 3, x ₁ is a failure of the CCU 102, x ₂ is a failure of the partner station 103, and y ₁ is a failure of the CCU 102 and the partner station 103. Connection is not possible (hereinafter referred to as connection impossible),
Let us consider a case where y ₂ is the text reception by the CCU 102 being unacceptable (hereinafter referred to as “reception not possible”), and y ₃ is the illegal text reception of the CCU 102 (hereinafter referred to as “receipt invalid”).

Here, the data flow from the observation of the failure phenomenon to the fuzzy diagnosis processor 104 is as follows. That is, the failure phenomenon data observed by the CCU 102 is transmitted to the diagnostic interface c, and every time the above y _i is observed by the CCU 102, the fuzzy diagnosis is performed as data digitized together with the total number of communications up to that point. Sent to the processor 104.

The fault data collection means 111 in the fuzzy diagnostic processor 104 sends the sent phenomenon data as fault data D of the CCU 102 to the processor 112 via the fault data interface d. It is assumed that the processor 112 updates the data of the total number of times of communication and the number of times of occurrence of each phenomenon and stores it as numerical data of the number of times of occurrence of each phenomenon / the total number of times of communication.

On the other hand, the membership function R is CCU1
It is assumed that the designer of No. 02 has defined it as in the equation 1 and that it is loaded from the membership function storage means 110 into the processor 112 using the function interface e.

[0026]

[Equation 1]

Elements of the membership function R, eg R ₁₁
And R ₁₂ are, when the unconnectable y ₁ is observed,
A failure of the CCU 102 has a causal relationship of 0.9, and a failure of the partner station 103 has a causal relationship of 0.6. Here, a communication abnormality is observed in the computer system, and the set Y of failure phenomena is the failure data D = [0.9 0.6
0.2] is recognized by the fuzzy diagnosis processor 104. This is the number of connections that cannot be made / total number of communications =
It is shown that 0.9, the unreceivable number / total number of communications = 0.6 and the incorrect reception number / total number of communications = 0.2.

Therefore, in the above Y = X * R,
When Y = D and the elements of the set X are x ₁ and x ₂ , the vector display is as shown in Equation ₂ . Note that 0 ≦ x ₁ ,
x ₂ ≦ 1.

[0029]

[Equation 2]

If the fuzzy operation * is an operation in which the equation 2 is expanded by a normal matrix operation and then the addition is rewritten as V and the multiplication is rewritten as Λ (generally called MAX-MIN combination), 0.9 = (0.9Λx ₁ ) V (0.6Λx ₂ ) 0.6 = (0.6Λx ₁ ) V (0.4Λx ₂ ) 0.2 = (0.2Λx ₁ ) V (0.5Λx ₂ ) Be done. Here, αΛβ is taken to be the smaller of α and β, and αVβ is taken to be the larger of α and β.

At, obviously (0.6Λ × ₂ )
Since <0.9, x ₁ ≧ 0.9.

In addition, obviously, 0.6 ≧
Since (0.4Λx ₂ ), x ₁ ≧ 0.6.

From the above, x ₁ ≧ 0.9 is obtained.

On the other hand, from the above, x ₁ ≧ 0.2 and 0.2 ≧
x ₂ is found, which is consistent with x ₁ ≧ 0.9.

From the above, 1 ≧ x ₁ ≧ 0.9 and 0.2
When ≧ x ₂ ≧ 0 and the failure data D is observed, the possibility of failure (x ₁ ) of the CCU 102 is 0.9 to
1 indicates that the other station 103 has a possibility of failure (x ₂ ) of 0 (no failure) to 0.2.

Here, in each j column of the membership function R _ij , the maximum value maxR _{ij of the} element or the minimum value minR _{ij of the} element and the corresponding j-th element y _j of the set Y of failure phenomena are If the relationship is y _j <minR _ij or y _j
When> maxR _ij , the equation of Y = X * R has no solution. Therefore, when y _j <minR _ij , y _j = mi
When nR _ij and y _j > maxR _ij , y _j = maxR _ij .

This corresponds to the case where, for example, [0.9 0.8 0.1] is obtained as the fault data D '. That is, focusing on the two column elements of the membership function R _ij , maxR _i2 = 0.6, and the failure data D ′
Element y ₂ = 0.8, the above y _j = maxR _ij
Therefore, it is shown that y ₂ = 0.6. Similarly,
For the three column elements of the membership function R _ij , y ₃ =
It has shown that it is set to 0.2.

The processor 112 performs the fuzzy operation * shown above, and the above operation result (interval value of x ₁ 1 ≧ x
_With respect to the interval value of ₁ ≧ 0.9 and x ₂ of 0.2 ≧ x ₂ ≧ 0), the elements x ₁ and x ₂ are used as actual device names or product names, and then sent to the display means 113 as analysis data. ..

The display means 113 is based on the analysis data (in this example, x ₁ : CCU 102, x ₂ : the partner station 10).
3) By displaying in a format that is easy for maintenance personnel to understand,
The cause of failure can be easily identified.

[0040]

As described above, according to the present invention, in the diagnostic processor used in the computer system, the membership function indicating the strength of the causal relationship between the expected cause of failure and the failure phenomenon is defined in advance, and actually By performing the fuzzy operation between the observed failure phenomenon and the membership function, it is possible to easily identify the cause of the failure.

[Brief description of drawings]

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

[Explanation of symbols]

 101 central processing unit (CPU) 102 communication control unit (CCU) 103 partner station 104 fuzzy diagnostic processor 110 membership function storage means 111 fault data collection means 112 processor (arithmetic processing unit) 113 display means a upper interface b communication line c diagnostics Interface d Fault data interface e Function interface f Display interface

Claims (1)

[Claims] 1. A diagnostic processor used in a computer system comprising a plurality of devices such as a central processing unit and a peripheral control unit, wherein a plurality of failure causes expected to occur in the computer system and the failure causes Membership function storing means for holding a membership function in which the strength of each causal relationship with a plurality of failure phenomena that are expected to be observed simultaneously or independently are numerically defined, and the computer system Fault data collecting means for recognizing a fault phenomenon that has occurred and collecting it as fault data represented by a numerical value, and inputting the fault data and the membership function into the fault data and the membership function based on fuzzy theory. On the other hand, a predetermined fuzzy calculation process is performed, and as a result of the calculation process, the failure Data processing unit that outputs the analysis data that numerically represents the strength of the causal relationship of each of the plurality of failure causes with respect to the data, and the strength of the causal relationship based on the analysis data by inputting the analysis data And a display means for displaying the fuzzy diagnostic processor.