JPH07262056A - Monitor device for failure occurrence trend of computer - Google Patents

Abstract

PURPOSE:To automatically specify a failure part and to analyze the future occurrence trend of the failure by utilizing a failure part specifying rule to specify the failure part based on the work recording. CONSTITUTION:A failure part specifying rule stored previously in a hard disk 3 is read out to a memory, and a pattern matching operation is carried out between the fault part specifying rule and the field data, and the failure part name is given to each item of the field data. A desired failure area to be analyzed is designated out of the examples of failure part selection screens shown on a CRT 4 and extracted from the memory. Then the monthly occurrence number of failures are calculated for each failure part name, and the accumulated number of failures are also calculated. Then, the number of working months and the accumulated number of failures are plotted on the abscissa and ordinate axes of a logarithmic graph respectively for each failure part based on the monthly occurrence number of failures, and a Weibull type hazard chart is produced. Then, a shape parameter (m), i.e., the tilt of a straight line approximate to the plot is decided, and the hazard chart and the value of the parameter (m), i.e., the processing results are shown on the screen of the CRT 4.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a new type of computer system, which is still in its infancy after the start of delivery, from a field work record including data and sentences concerning a phenomenon, cause, and countermeasure of a failure created by a maintenance worker. The present invention relates to a failure occurrence tendency monitoring apparatus for a computer that analyzes future failure occurrence tendency for this computer system, creates a hazard Weibull chart, and calculates and displays the slope m thereof.

[0002]

2. Description of the Related Art After a certain time has passed since the start of delivery of a computer system, the initial failure, which is often the case with new models, is usually calmed down and stabilized, becoming the mother of fault data collection. The number of deliveries has increased, and the tendency for failures to occur has converged to a certain degree and stabilized. If you reach such a stage, code the phenomenon name and cause name of the failure,
Using the conventional EDP technology, it becomes possible to analyze the fault by classifying the code. However, as long as the new type of computer system has not yet passed after the start of delivery and the failure tendency is not stable,
A failure phenomenon created by a computer maintenance worker when a failure occurs in the computer system and countermeasures are taken, or when inspection is performed for preventive maintenance even if the customer has not received any complaints about the failure. It is clear that the field work record including data and texts regarding causes, countermeasures, potential cause events of failure occurrence, etc. will be valuable materials for analyzing and predicting future failure occurrence tendency for the system.

Hitherto, a skilled person has deciphered this field work record and analyzed the fault. That is, the failure data is edited to obtain the failure rate of the computer system, dots are plotted on the Weibull type hazard chart to check the failure occurrence tendency, and the maintenance time data is edited to calculate the mean failure time (MDT) of the computer system. : Mean Down Tim
I have conducted human analyses, such as estimating e), such as reliability and maintainability. However, the number of engineers who can perform such an analysis is limited, and there is a problem in that this work requires a considerable number of man-hours.

[0004]

It is a great burden for the maintenance work manager to manually decipher a huge field work record by the maintenance worker in order to analyze the tendency of future failure of a computer system or the like. However, it is conventionally considered that mistakes and leaks easily occur, and various countermeasures have been proposed.

As a hardware failure analysis method of this kind, for example, the field data is converted into a bar code and read by a bar code reader, the field data is accumulated, the field data is edited and analyzed, and the analysis result is displayed. The technique of displaying in graph form has been disclosed as a reliability analysis system in Japanese Patent Laid-Open No. 4-172553. However, although this technology helps to make reading easier, it will not be accurate if you try to apply it to a computer system that has not been known for a long time after the start of delivery and the phenomenon, cause, etc. of the failure occurrence are unknown. Even if the classification code is not defined yet, or even if it is defined assuming various cases, it is rare that they accurately correspond to the actual failure, and in the end, it is edited and analyzed. The result will be unreliable. In addition, JP-A-2-81299
In the gazette, variable failure judgment levels are input for periodic inspection data of products such as measuring instruments, accumulated data such as error data, and hazard and Weibull charts are created according to the failure judgment levels. There is disclosed a technology of a failure prediction device used for calculating the inclination m and determining a potential failure state such as an initial failure, a random failure, and a wear failure of a product.
This technology is also effective if it is applied to so-called withered products that have already been used as product types, but it is not suitable for monitoring the failure occurrence tendency of new types of computer systems.

The present invention automatically and mechanically reads a field work record created by a maintenance worker of a computer system for a new type of computer system, which is still new after the start of delivery, to almost automatically detect a faulty part. It is an object of the present invention to provide a failure occurrence tendency monitoring device for a computer, which is specified by a specific method and analyzes the future failure occurrence tendency from the data thus accumulated.

[0007]

In order to solve the above problems, according to the present invention, text data relating to a phenomenon, cause, and countermeasure of a fault created by a computer maintenance worker after finishing the fault countermeasure work or preventive maintenance work. A means for automatically identifying a faulty part from a field work record including the, a means for accumulating the number of cases for each faulty part found by this means, and an editing process for editing the data concerning the faulty part accumulated by this means Means and a failure occurrence tendency analysis means for analyzing the future failure occurrence tendency for each specific failure area based on the data concerning the failure area edited by this editing means and displaying the analysis result as a chart. It is provided.

[0008]

In order to automatically identify the faulty part from the field work record, the present inventors have stated that "a method of creating a knowledge base for maintenance of an information processing device and a classification device based on the maintenance knowledge base of a maintenance work report" are used. Already, Japanese Patent Application No. 4-3382
Proposed by No. 49. Since field work records are currently created by word processors, they are easy to read through OCR. As described in detail in the specification of the above-mentioned patent application, in the field maintenance work record, technical terms related to various maintenance techniques, for example, related to phenomena, causes, countermeasures, parts, etc. of failures frequently appear. If these are stored in the computer used for the maintenance work in advance as keywords relating to the maintenance technique, a maintenance knowledge base for the computer to be the maintenance work can be created. In the present invention, the content of the above proposal is used as a faulty part identification rule. Utilizing this fault location identification rule, the content of the maintenance work report is edited by identifying the fault location from the field work record, and the future failure occurrence tendency for each specific failure location is analyzed, and the analysis result is displayed. Make it possible to display it as a graph. For example, as a result of the above analysis, a Weibull-type hazard chart is displayed, and a failure occurrence tendency can be determined by a shape parameter obtained from the Weibull-type hazard chart. That is, when the inclination (shape parameter m) of the Weibull-type hazard chart is larger than 1, the failure rate increases in the failure monitoring period, and when it is smaller than 1, the failure rate decreases, and when it is 1, the failure rate decreases. Indicates that the failure rate is stable.

[0009]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A computer fault occurrence tendency monitoring apparatus according to the present invention will be described below in more detail with reference to the drawings. It should be noted that the computer failure occurrence tendency monitoring device embodying the present invention is not usually installed as a single device independently, but rather in a computer system targeted for failure occurrence tendency monitoring. As shown in FIG. 3, the input device, the display device, the external storage device, the printer, and the like, which are built in and are commonly used, are also shared with the computer that is the target of the failure occurrence tendency monitoring. That is, FIG. 2 is a schematic configuration diagram of a data processing device (computer system) which is capable of monitoring and analyzing a failure occurrence tendency by implementing the present invention. In FIG. 2, reference numeral 1 denotes an arithmetic processing unit having a fault occurrence monitoring function according to the present invention, which comprises a CPU, a memory and the like. 2 is an input device for inputting field data,
It is equipped with a keyboard and mouse. Reference numeral 3 is a hard disk in which rules for identifying a failed portion are stored in advance, 4 is a portion (unit, etc.) desired to be analyzed,
A CRT (cathode ray tube) 5 for displaying a hazard chart is a printer for outputting a Weibull type hazard chart or the like as a result of the arithmetic processing.

FIG. 1 is a flow chart for explaining the processing flow of an embodiment of the failure occurrence tendency monitoring apparatus of the present invention. The processing performed in each frame in FIG. 1 will be sequentially described.
First, in process 100, the device name, failure date, phenomenon name, etc. of the field data example shown in FIG. 3 are input by the input device 2 shown in FIG. Next, in process 101, the input value is stored in the hard disk 3. OCR for these processes
Can be utilized to reduce man-hours. Then process 1
In 02, the fault site identification rule shown in FIG. 4, which is stored in the hard disk 3 in advance, is read out to the memory. Then, in process 103, pattern matching is performed between the field data in the hard disk and the failure site identification rule, and a failure site name is added to each piece of field data as in the example shown in FIG. go. Next, in process 104, the failure site (unit or the like) to be analyzed is designated by the mouse of the input device 2 from the failure site selection screen example displayed on the CRT 4 as shown in FIG. Then, in process 105, the field data for the designated failure site is extracted from the memory, the monthly failure count is calculated for each failure site, and the cumulative failure count is also calculated. An example thereof is shown in FIG. Next, in process 106,
Based on the number of failures per month for each failure location, the horizontal axis of the logarithmic graph plots the number of operating months on the horizontal axis and the cumulative failure number on the vertical axis. The shape parameter m, which is the inclination, is determined. The hazard chart and the value of m, which are the results of this processing, are displayed on the screen of the CRT 4. In this calculation process, originally the cumulative hazard (the value obtained by dividing the number of failures by the number of operating units) should be plotted on the vertical axis. Therefore, in the above arithmetic processing, the cumulative number of failures is plotted on the vertical axis on the assumption that the number of operating machines is small (no change) within the target operation monitoring period, for example, several months. Is a feature. By doing so, the processing becomes simple. Then, in process 107, the value of the shape parameter is output. Then, the process 108
Then, the failure data of the base to be analyzed is output.

FIG. 8 shows an output example of the Weibull type hazard analysis of the field data shown in FIG. In the case of this example, since the shape parameter m is 1.1, it is known that the number of failures tends to increase during this period, suggesting that preventive maintenance is required.

[0012]

As described above, according to the present invention, a field work record including text data relating to a phenomenon, cause, and countermeasure of a fault created by a computer maintenance worker after finishing the fault countermeasure work or preventive maintenance work. Therefore, it becomes possible to monitor the future failure occurrence tendency for each part (unit, etc.) almost automatically with an extremely small number of man-hours, and to understand common causes of failure, so preventive maintenance plans can be made. To be

[Brief description of drawings]

FIG. 1 is a flow chart for explaining a flow of processing of an embodiment of a failure occurrence tendency monitoring device of the present invention.

FIG. 2 is a schematic configuration diagram of a data processing device (computer system) that is capable of monitoring and analyzing a failure occurrence tendency by implementing the present invention.

FIG. 3 is a diagram showing a display screen of a CRT displaying a device name, a failure occurrence date, a phenomenon name, etc. as an example of field data.

FIG. 4 is a diagram showing an example of a fault site identification rule created based on the prior application of the present inventors.

FIG. 5 is a diagram showing a state in which a failure part identified by a failure part identification rule is added and displayed for each item of the field data shown in FIG.

FIG. 6 is a diagram showing an example of a failure site selection screen displayed on a CRT for designating a site desired to be analyzed for a failure occurrence tendency with a mouse or the like.

FIG. 7 is a diagram showing a state in which field data for a designated failure part is extracted from a memory and a monthly failure number and a cumulative failure number are calculated for each failure part.

8 is a diagram showing an example in which the field data shown in FIG. 7 is displayed as a Weibull-type hazard chart, showing that the number of failures tends to increase during the monitoring period with the shape parameter m = 1.1. is there.

[Explanation of symbols]

1 ... Arithmetic processing device having a failure occurrence tendency monitoring function 2 ... Input device for inputting field data etc. 3 ... Hard disk storing rules for specifying failure site 4 ... Analysis output for selecting a desired failure site CRT 5 ... Printer that outputs Weibull-type hazard charts of calculation results 100 ... Field data input processing 101 ... Hard disk storage processing 102 ... Hard disk failure site identification rule reading processing into memory 103 ... Hard disk phenomenon Processing for classifying sentences of names according to the rule for identifying a failure site and storing in memory 104 ... Processing for designating a desired site for analysis with a mouse 105 ... Extracting and classifying and editing data in a memory for a desired site for analysis designated by a mouse Process 106: Weibull type hazard chart creation Calculation processing 107 ... failure content output process corresponding to the specified analysis desired site in the processing 108 ... mouse outputs the value of the shape parameter m of the Weibull chart

Claims (2)

[Claims] 1. A means for automatically identifying a failure portion from a field work record including text data relating to a phenomenon, cause, and countermeasure of a failure created when a maintenance worker of a computer finishes a failure countermeasure work or preventive maintenance work. , A means for accumulating the number of cases for each faulty part found by this means, an editing processing means for editing data on the faulty part accumulated by this means, and data on the faulty part edited by this editing means. A failure occurrence tendency monitoring device for a computer, characterized by further comprising failure occurrence tendency analysis means for analyzing a future failure occurrence tendency for each specific failure part and displaying the analysis result as a chart. . 2. The method according to claim 1, wherein when analyzing the failure occurrence tendency, the number of operating machines within the monitoring period is considered to be small, and the cumulative number of failures is used instead of the cumulative hazard. Computer fault occurrence trend monitoring device.