WO1995022089A1

WO1995022089A1 - Method and apparatus for diagnosing fault

Info

Publication number: WO1995022089A1
Application number: PCT/JP1995/000171
Authority: WO
Inventors: Hiroyoshi Yamaguchi
Original assignee: Komatsu Ltd.
Priority date: 1994-02-09
Filing date: 1995-02-08
Publication date: 1995-08-17
Also published as: JPH07225610A

Abstract

This invention relates to a method and an apparatus for diagnosing a fault quickly and accurately. Various inspection items for fault diagnosis are classified into sensor inspection items and input data inspection items. As to the sensor inspection items, the degree of abnormality is computed on the basis of a detection value of a sensor. As to the input data inspection items, the degree of abnormality is acquired by data input. The detection value of each sensor and the abnormality judgement result of each sensor are displayed during the fault diagnosing process, and the detection values of the sensors can be appropriately changed and modified from the display content. The degree of abnormality is displayed for each sensor inspection item computed, and the degree of abnormality can be appropriately changed and corrected from the display content. The degree of reliability is computed for each of the various causes of fault on the basis of the data representing the degree of association of various inspection items and various causes of fault and the data of the degree of abnormality obtained finally.

Description

Description Failure diagnosis method and device Technical field

The present invention relates to a method and an apparatus for diagnosing a failure of a failure diagnosis target such as a construction machine. Background art

There are two types of conventional fault diagnosis systems:

A. Offline diagnostic system

This system asks the operator on the screen about the items to be diagnosed at the time of failure diagnosis on the screen, and the operator inspects these inspection items, and inputs the inspection results as abnormalities for each inspection item. It uses this data to perform fault diagnosis, also called an interactive diagnosis system:

B. Online diagnostic system

This system captures sensor signals from sensors installed at the target of failure diagnosis at the time of failure diagnosis, calculates the degree of abnormality based on these sensor signals, and performs failure diagnosis using the calculated degree of abnormality. It is also called an automatic diagnostic system-

C. Online type abnormality monitoring system

This system always captures sensor signals from sensors arranged for failure diagnosis, determines whether or not there is an abnormality based on these sensor signals, and issues a warning if it is determined to be abnormal. It emits:

The abnormality monitoring system described in C above simply determines the abnormality at the location detected by the sensor, and cannot identify the cause of the failure:

In this regard, the diagnostic systems A and B identify the cause of the failure, but the off-line diagnosis of A has the problem that it takes time for the operator to perform all inspections: The degree of abnormality entered as data Another problem is that depending on the subjectivity, different failure causes are diagnosed by different people.

On the other hand, in the online diagnosis of B, the inspection items that cannot be detected by the sensor cannot be used for failure diagnosis: Therefore, the failure diagnosis is performed based on only the limited inspection items. There is a problem that the accuracy of the diagnosis is deteriorated. Also, when an output from the sensor cannot be obtained due to a malfunction such as a sensor failure, or when the sensor outputs a clearly abnormal value. However, the accuracy of the fault diagnosis is reduced:

The present invention has been made in view of such circumstances, and can solve the problem of performing a failure diagnosis only by the online type diagnosis or only the offline type diagnosis, and can obtain an accurate diagnosis result in a short time. It is an object of the present invention to provide a failure diagnosis method and apparatus which can be performed.

Therefore, in the first invention of the present invention, a failure for diagnosing the failure of the failure diagnosis target based on the relevance data indicating the degree of association between the various inspection items of the failure diagnosis target and the various failure causes of the failure diagnosis target. In the diagnostic method,

The relevance data obtained by classifying the various inspection items into a sensor inspection item inspected by a detection value of a sensor disposed on the failure diagnosis target and an input data inspection item inspected by input data. And preparing a failure to be diagnosed based on the relevance data.

Also, in the second invention of the present invention, a knowledge base having relevance data indicating the degree of relevance between various inspection items to be diagnosed and various causes of the failure to be diagnosed is provided. A fault for inferring a certainty factor for each cause of failure based on the abnormality degree data indicating the degree of abnormality of various inspection items and the association degree data, and diagnosing the failure to be diagnosed based on the inference result. In a diagnostic device,

The various inspection items are classified into a sensor inspection item inspected by a detection value of a sensor disposed on the failure diagnosis target and an input data inspection item inspected by input data, and the knowledge base is classified. Create a resource in advance, A threshold value is set for each sensor, and a process of determining whether or not the sensor is abnormal is performed by comparing the set threshold value with a detection value of the sensor.

In addition to displaying the detection value of each sensor, displaying the abnormality determination content for each sensor, and performing an operation of changing the detection value of each sensor based on the display content,

Based on the detection value of each sensor subjected to the change operation, perform a process of calculating the degree of abnormality for each sensor inspection item,

Displays the calculated degree of abnormality of each sensor inspection item, performs an operation to change the degree of abnormality of each sensor inspection item based on the display contents, and inputs necessary abnormality degree data for each input data inspection item. Do

Based on the degree of abnormality of each sensor check item subjected to the change operation and the degree of abnormality of each input data check item subjected to the data input operation, the degree of certainty for each cause of failure is inferred, and based on the inference result, The failure diagnosis target is diagnosed.

According to the configuration of the first aspect, the inspection items of the sensor can be quickly inspected based on the detection value of the sensor without causing individual differences: Further, the degree of abnormality is obtained from the detection value of the sensor. Inspection items that cannot be checked can be inspected by inputting data as input data inspection items, and accurate inspection results can be obtained with a large number of inspection items that are not limited to sensor inspection items: According to the configuration of the second invention, similarly to the first invention, the degree of abnormality of the sensor inspection item can be quickly obtained based on the detection value of the sensor without causing individual differences. For inspection items for which the degree of abnormality cannot be determined from sensor detection values, the degree of abnormality can be obtained by inputting data as input data inspection items. High accuracy and fault diagnosis results can be obtained:

In addition to this, the detected value of each sensor and the result of abnormality determination of each sensor are displayed, and the detected value of the sensor can be changed and corrected as appropriate from the displayed contents, and the degree of abnormality for the calculated sensor inspection item is displayed. , The degree of abnormality can be changed and corrected as appropriate from the displayed content: For this reason, even if there is an abnormality such as a failure of the sensor itself or an error when calculating the Can determine abnormalities Since the display contents can be corrected, accurate fault diagnosis results can be obtained without being based on abnormal data.

Figure 1 is a fault diagnosis method and device flow Chiya one preparative illustrating a processing procedure of the embodiment of the present invention _c

FIG. 2 is a flowchart illustrating a processing procedure of another embodiment of the failure diagnosis method and apparatus according to the present invention.

FIG. 3 is a diagram showing the contents of a knowledge base for failure diagnosis applied to the embodiment.

FIG. 4 is a diagram showing an abnormality degree display screen displayed on the display screen in the process of failure diagnosis.

FIG. 5 is a diagram illustrating a membership function for calculating the degree of abnormality.

FIG. 6 is a diagram showing a sensor value display screen displayed on the display screen in the process of failure diagnosis. FIG. 7 is a diagram showing a diagnosis result display screen showing the result of the failure diagnosis. Best form for

Hereinafter, embodiments of a failure diagnosis method and apparatus according to the present invention will be described with reference to the drawings. In the embodiment, it is assumed that a failure diagnosis of a construction machine is performed: Specifically, a personal computer (hereinafter referred to as a personal computer) performs a failure diagnosis of a construction machine according to the procedure shown in FIG. The knowledge base 1 for fault diagnosis will now be described with reference to FIG. 3. As shown in FIG. 3, the structure of the knowledge base 1 is basically based on the inspection item i of the construction machine on the line side, that is, “ The battery voltage (low), the exhaust temperature (low), ... and the failure cause j of the row-side construction machine, that is, ^Γbattery failure, injection nozzle failure, electric heater failure, etc. It has a matrix structure, and the degree of association Wi j indicating the degree of association between these various inspection items i and various failure causes j is preset and stored as data: The degree of association Wij is given as a numerical value in the range of 0≤Wi j≤1, and the degree of association Wi j = 0 corresponds to "no association", and the degree of association Wi j = l is "maximum association". There is a corresponding to:

Various inspection items i include sensor inspection items that are inspected based on the detection values of sensors installed on construction machinery, and input data inspection items that cannot be inspected based on the detection values of sensors (inspection by operators). Items that can be inspected based on the values detected by the sensors may be included in the input data inspection items:

Such a knowledge base 1 is set and stored for each fault condition of the construction machine, that is, for “bad engine startability”:

In addition, a sensor group corresponding to the above-mentioned various sensor check items, that is, a battery voltage sensor for detecting a battery voltage, an exhaust temperature sensor for detecting an exhaust temperature, and the like, are naturally provided at a predetermined location on the construction machine before failure diagnosis. These sensors and the personal computer should be wired so that sensor signals can be input to the computer via a predetermined interface-as shown in Fig. 1. First, when performing a failure diagnosis, the operator operates the keyboard of a personal computer, for example, to check the current failure state of the construction machine to be diagnosed, for example, “Fault (1): Poor engine startability. Is input as data (step 101):

Then, each sensor corresponding to the various sensor inspection items shown in the read knowledge base 1 is selected, and the input of the sensor signal is controlled so that the detection value of each of the selected sensors is obtained. As a result, the detected values of the battery voltage sensor, exhaust temperature sensor, etc. corresponding to the sensor inspection items “battery voltage”, “exhaust temperature”, etc. shown in FIG. 3 are input and acquired in the personal computer (step 1 0 2)

Next, it is determined whether or not the sensor is abnormal based on the detection signal of each sensor taken into the personal computer:

That is, the threshold value of the output is set in advance for each sensor, and whether or not the sensor is abnormal is determined based on whether or not the threshold value is equal to or higher than the threshold value.

For example, for an exhaust temperature sensor, a first threshold of "0 ° C" A second threshold value of “0 0 0 ° C” is set: Therefore, if the “detected value (exhaust gas temperature) is less than or equal to the first threshold value (0 ° C), (Temperature) is greater than or equal to the second threshold value (100 ° C), the sensor (exhaust temperature sensor) is abnormal: '' is determined in advance.

Therefore, the detection value of the exhaust gas temperature sensor taken in step 102 is applied to the abnormality determination rule to determine whether the exhaust gas temperature sensor is abnormal (step 103);

Next, the detection values of each sensor captured in step 102 are displayed on the display screen of the display unit, and the sensor abnormality determination result in step 103 described above corresponding to each sensor detection value is displayed. Is also displayed:

That is, as shown as a sensor value display screen 3 in FIG. 6, for each type of sensor such as battery voltage, exhaust temperature, lubricating oil pressure, and the like, the detected values are converted into predetermined engineering units and displayed. For example, the current battery voltage is 12 V, and the lubricating oil pressure is 20 kg Z cm2. Then, depending on the detection value of each sensor, the sensor abnormality judgment result is “normal” or “normal”. Abnormal "is displayed-For example, the battery voltage is 12 V and" Normal ", the lubricating oil pressure is 2111 and the normal is as follows:

However, as described above, for example, if the sensor signal of the exhaust temperature sensor is determined to be abnormal as a result of the abnormality determination according to the abnormality determination rule described above, the detected value of the sensor is displayed on the sensor value display screen 3. Not:

In this case, since the true exhaust temperature is unknown due to a sensor failure, etc., the exhaust temperature obtained separately by the predetermined detection means or the exhaust temperature determined based on the operator's experience etc. is keyboard. As a result, the values for all sensors corresponding to each sensor inspection item are accurately given, and the sensor values within these normal ranges are indicated by broken lines on the sensor value display screen 3. Will be displayed on the sensor value display section 3a as shown in (step 104):

Next, based on a predetermined abnormality degree evaluation function, the abnormality degree I i is calculated for each sensor inspection item i: Here, the abnormality degree I i is the abnormality degree of the inspection item i. Is given as a numerical value in the range of 0≤I i≤1, and the degree of abnormality I i = 0 is Abnormality I i = l corresponds to “the largest abnormality”.

For example, taking “exhaust gas temperature” as an example of a sensor check item, a membership function that is an abnormality evaluation function as shown in Fig. 5 is prepared: Therefore, the final decision in step 104 was made. The exhaust gas temperature is applied to this membership function to calculate the degree of abnormality I i for the inspection item “exhaust gas temperature”. That is, assuming that the exhaust gas temperature is currently 50 °, the arrow in FIG. As shown in (1), the degree of abnormality Ii indicating that the exhaust gas temperature is low due to the membership function is determined to be 0.8 (step 105).

Next, as shown in FIG. 4, the result of the calculation of the degree of abnormality in the above step 105 is displayed on the display screen as the surface 2 of the degree of abnormality display:

That is, as shown in FIG. 4, the abnormality level Ii is displayed for each of the sensor check items i: "low battery voltage", "low exhaust temperature", "high lubrication oil pressure": For example, as described above, regarding “exhaust gas temperature”, the degree of abnormality “0.8” is displayed corresponding to the check item “exhaust gas temperature is low”. Similarly to the above, for the degree of abnormality I i, which could not be calculated due to an error such as an error in the calculation, the operator can perform a keyboard operation based on experience or the like to change the value. In addition, even if the calculated degree of abnormality I i is determined to be abnormal in view of the failure state, the value can be similarly determined by operating the keyboard. Can be modified.

On the other hand, the abnormality level display screen 2 also displays each input data inspection item, and for the abnormality degree corresponding to these input data inspection items, the value determined by the operator based on his / her own experience and the like is operated by keyboard operation. As a result, all the abnormalities Ii corresponding to each sensor inspection item are given as those within the appropriate range, and all the abnormalities corresponding to each input data inspection item are obtained. I i is given as being within the proper range, and these abnormalities I i will be displayed on the abnormal degree display section 2 a shown by the broken line on the abnormal degree display screen 2 (step 106). Based on the knowledge base 1 shown in Fig. 3, the fault diagnosis is performed: The failure diagnosis is performed by calculating the certainty factor CFj of various failure causes j based on the relevance data Wij of the knowledge base 1 and the abnormality level Ii of each inspection item i obtained in step 106 above. is there:

For example, the following equation (1) can be used to calculate the confidence factor CFj:

CFj = ∑Wij * I i

i

(Step 107) Next, the calculation result of the above step 107 is displayed on the diagnostic result display screen 4 as shown in FIG. 7: That is, as shown in FIG. , Various failure causes J, that is, “battery failure”, “injection nozzle failure” ... are displayed in units of “25%”, “89%” ... When a predetermined cause of failure is selected in the failure cause display section 4a, a preset troubleshooting method is displayed on the screen corresponding to the selected failure cause = For example, the failure cause `` Battery failure Is selected, a detailed explanation of how to check for a battery failure and how to take measures against the failure is displayed on the screen, based on which the operator can take immediate action (step 108),

FIG. 2 shows a failure diagnosis processing procedure according to another embodiment. The difference from the flow chart shown in FIG. 1 described above is that the failure input processing corresponding to step 101 in FIG. There is no corresponding action:

In this embodiment, in steps 201 to 207, the same processing as in steps 102 to 108 in FIG. 1 is executed. That is, in this embodiment, the necessary knowledge base 1 is not extracted in accordance with the failure state, and the process proceeds based on the general-purpose knowledge base 11 which does not specify the failure state. In step 205, the data from all input data check items will be input as well as the detection values from the sensors corresponding to the check items.

As described above, according to the present invention, various inspection items for performing failure diagnosis are provided. Sensor inspection items and input data items are classified, so sensor inspection items can be inspected quickly without individual differences based on sensor detection values, and are not limited to sensor inspection items Accurate failure diagnosis results can be obtained by many inspection items

Further, according to the present invention, in the course of the failure diagnosis, the detection value of each sensor and the abnormality determination result of each sensor are displayed, and the detection value of the sensor can be appropriately changed and corrected based on the display contents, The abnormalities of the detected sensor inspection items are displayed, and the abnormalities can be appropriately changed and corrected based on the display contents. Therefore, a failure occurs in the sensor itself, or an error occurs when the abnormalities are calculated. Even if there is an abnormality such as an error, it is possible to determine and correct the abnormality based on the displayed content, and to obtain an accurate failure diagnosis result based on no abnormal data.

Claims

The scope of the claims

1. A failure diagnosis method for diagnosing a failure of a failure diagnosis target based on association degree data indicating a degree of association between various inspection items of the failure diagnosis target and various failure causes of the failure diagnosis target,

The relevance data obtained by classifying the various inspection items into a sensor inspection item inspected by a detection value of a sensor disposed on the failure diagnosis target and an input data inspection item inspected by input data. A failure diagnosis method for diagnosing the failure to be diagnosed based on the relevance data.

2. A failure diagnosis device that diagnoses a failure of the failure diagnosis target based on association degree data indicating a degree of association between various inspection items of the failure diagnosis target and various failure causes of the failure diagnosis target,

The various inspection items are classified into a sensor inspection item inspected by a detection value of a sensor arranged in the failure diagnosis target and an input data inspection item inspected by input data. A failure diagnosis apparatus that creates a knowledge base having relevance data indicating the degree of relevance between various inspection items and the various failure causes, and diagnoses the failure to be diagnosed based on the knowledge base.

3. Prepare a knowledge base that contains the degree of association between the various inspection items to be diagnosed and the various causes of the failure to be diagnosed, and use the abnormality degree data that indicates the degree of abnormality of the various inspection items. A failure diagnostic device for inferring certainty factors for various failure causes based on the inference data and the relevance data, and diagnosing the failure to be diagnosed based on the inference result;

The various inspection items are classified into a sensor inspection item inspected by a detection value of a sensor disposed on the failure diagnosis target and an input data inspection item inspected by input data, and the knowledge base is classified. Create a resource in advance,

A threshold value is set for each sensor, and a process of determining whether or not the sensor is abnormal is performed by comparing the set threshold value with a detection value of the sensor.

In addition to displaying the detection value of each sensor, the details of abnormality determination for each sensor are also displayed. Performing an operation of changing the detection value of each sensor based on the display content;

Based on the degree of abnormality of each sensor check item subjected to the change operation and the degree of abnormality of each input data check item subjected to the data input operation, a certainty factor for each cause of failure is inferred.

A failure diagnosis device for performing a failure diagnosis on the failure diagnosis target based on the inference result:

4. The knowledge base is prepared in advance for each failure state of the failure diagnosis target,

When data indicating a failure state is input, a knowledge base corresponding to the input data is selected,

While controlling to input the detection value of each sensor corresponding to each sensor check item indicated in the selected knowledge base, displaying the content of the selected knowledge base, and performing each input based on the display content 4. The failure diagnosis device according to claim 3, wherein the abnormality degree data corresponding to the data check item is input and operated.