JP2016115008A - Failure prediction device, failure prediction method and failure prediction program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To calculate a generation probability of failure with consideration of a deterioration state.SOLUTION: A first derivation part 11 derives a function expression, which expresses a failure ratio of each steel pole at each use time based on inspection information, with the use time t as a variable number, from all records of the inspection information. A use time estimation part 12 uses the inspection information as input, and estimates an average use time which is a period until the respective steel pole is deteriorated to respective deterioration state. A second derivation part 13 adds the average use time which is the period until the respective steel pole is deteriorated to the respective deterioration state, estimated by the use time estimation part 12, to the time t which is the variable number of the function expression indicating the failure ratio of the respective steel pole at each use time, derived by the first derivation part 11. A future failure ratio calculation part 14 uses a use year and the deterioration state at a future prediction time T1 as inputs to a function expression of an average failure ratio curve which is derived by the second derivation part 13, and calculates the predicted failure ratio of the respective steel pole according to the use year and deterioration state.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a failure prediction apparatus, a failure prediction method, and a failure prediction program.

  The modern society is supported by various facilities and structures. Equipment, structures and the like deteriorate with various factors over time and lose their functions. Equipment and structures are managed based on appropriate maintenance and renewal plans. For example, in the case of facilities and structures that have a great social impact when deterioration progresses and damage or failure occurs, preventive maintenance based on the prediction of deterioration before damage or failure is possible so that damage or failure can be avoided as much as possible. Is desired to be implemented.

  As one of the methods of predicting deterioration, there is a method of obtaining a failure rate curve function equation according to the Weibull distribution by Weibull analysis, and predicting failure rates of individual facilities and structures using the failure rate curve function equation. is there. Here, the Weibull distribution is that when W. Weibull conducted a study on the fatigue life of metallic materials in 1939, “the failure of brittle materials is the presence of mutually independent defects distributed in the material. It is a model proposed based on the assumption that "is caused by". Since the Weibull distribution can also be applied to the "model where the entire system cannot be used due to a failure of the most vulnerable part of the system" that frequently appears in reliability engineering, it is adopted as a probability distribution model of the occurrence probability of failures. The

  In the Weibull distribution, the failure rate is defined by the following (Equation 1).

  At this time, m is called a shape parameter, and η is called a scale parameter (or characteristic lifetime), both of which are Weibull distribution parameters. Determining both parameters is equivalent to determining the failure rate. If both parameters are determined, the future failure rate can be predicted only by the usage time (or usage frequency) t.

Kenji Tanaka, “Introductory reliability engineer learns for the first time”, Nikka Giren Publisher, December 25, 2008 Masahiro Tokuma, Takashi Sawada, “Reliability analysis of submerged corrosion of steel pipe columns”, Proceedings of the 25th Autumn Reliability Symposium of the Japan Reliability Society, pp.21-24, November 2, 2012

  However, the equipment and structures used in the actual environment are affected by various environmental factors, and the deterioration state is not necessarily the same even when the usage time is the same at the latest equipment inspection. For example, equipment installed outdoors is affected by weather factors. That is, at the latest equipment inspection, even if the equipment and structure have the same usage time, the more serious the deterioration state, the higher the probability of occurrence of a failure. Thus, in order to perform preventive maintenance in consideration of the deterioration state at the time of the latest equipment inspection, it is desired to calculate the probability of occurrence of a failure in consideration of the deterioration state.

  An example of an embodiment disclosed in the present application has been made in view of the above, and an object thereof is to calculate the occurrence probability of a failure in consideration of a deterioration state.

  An example of an embodiment disclosed in the present application includes a first derivation unit, an estimation unit, a second derivation unit, and a calculation unit. The first deriving unit predicts an average failure rate for a certain usage time of the management target object from data obtained by accumulating inspection results of each management target object that are classified into three or more deterioration states in stages. Deriving the function formula of the rate curve. The estimation unit estimates each average usage time until each managed object reaches each degradation state. The second deriving unit corrects the functional equation of the average failure rate curve derived by the first deriving unit with each average usage time estimated by the estimating unit, thereby obtaining each of the deterioration states. A function formula of a failure rate curve that predicts a failure rate in a certain usage time of a management object is derived. The calculation unit calculates each failure rate at which each managed object corresponding to each deterioration state fails during a certain usage time using the functional expression of the failure rate curve derived by the second deriving unit.

  According to an example of an embodiment disclosed in the present application, for example, it is possible to calculate the occurrence probability of a failure in consideration of a deterioration state.

FIG. 1 is a diagram illustrating an example of the configuration of the failure prediction apparatus. FIG. 2 is a diagram illustrating an example of inspection information data. FIG. 3 is a diagram illustrating an example of details of processing of each unit of the failure prediction apparatus. FIG. 4 is a diagram illustrating an example of a Weibull plot when the degradation state 1 is a failure. FIG. 5 is a diagram illustrating an example of a Weibull plot when the degradation states 1 and 2 are failures. FIG. 6 is a diagram illustrating an example of a Weibull plot when the degradation states 1 to 3 are failures. FIG. 7 is a diagram illustrating an example of each parameter. FIG. 8 is a flowchart showing failure prediction processing. FIG. 9A is a diagram showing a ranking of failure rates calculated by the conventional technique. FIG. 9B is a diagram illustrating a ranking of failure rates calculated by the technique of the embodiment. FIG. 10 is a diagram illustrating an example of a computer in which the failure prediction apparatus is realized by executing each program.

[Embodiment]
Hereinafter, embodiments of a failure prediction device, a failure prediction method, and a failure prediction program disclosed in the present application will be described. In the following embodiments, steel pipe columns installed outdoors are targets for failure rate prediction. In other words, the embodiment simplifies the difference in the deterioration state derived from the difference in the use environment that cannot be grasped only by the actual use time when determining the priority order of the steel pipe column preventive maintenance, for example, the equipment renewal or the equipment inspection. We present a failure prediction method that can be considered statistically. However, management targets for failure rate prediction are equipment, facilities, etc. that have deteriorated due to factors other than aging, such as the installation environment, and that have failed or require replacement. What is necessary is just to install many things under various geographical conditions or weather conditions. The following embodiments are merely examples, and do not limit the technology disclosed by the present application. Moreover, you may combine suitably embodiment shown below and its modification in the range which does not contradict.

(Configuration of failure prediction device)
FIG. 1 is a diagram illustrating an example of the configuration of the failure prediction apparatus. The failure prediction device 10 is connected to the inspection information database 1 and the output device 2. The inspection information database 1 has a record in which at least three pieces of deterioration information indicating the degree of progress of deterioration in at least three stages, the identification information of the steel pipe column, the construction date, and the latest inspection date are associated with each steel pipe column. The inspection information database 1 stores a sufficient number of records (data number). The output device 2 is an output device such as a display or a printer that outputs a failure prediction result from the failure prediction device 10.

  The failure prediction apparatus 10 includes a first derivation unit 11, a usage time estimation unit 12, a second derivation unit 13, and a future failure rate calculation unit 14. The first deriving unit 11 receives the inspection information from the inspection information database 1, and from all records of the inspection information, the failure rate at each usage time of each steel pipe column based on the inspection information, and the usage time t as a variable. A function expression expressed by the function to be derived is derived.

  The usage time estimation unit 12 receives inspection information from the inspection information database 1 as input, and estimates an average usage time at which the steel pipe column deteriorates to each deterioration state. In other words, the usage time estimation unit 12 receives the inspection information as input, and regards all deterioration states indicating deterioration that has progressed to each deterioration state or more among all records of the inspection information as failure states. And the usage time estimation part 12 considers all the deterioration states which show the deterioration which has not progressed to each deterioration state as a non-failure state. A deterioration state indicating deterioration that has progressed to each deterioration state or more is referred to as a failure state, and a deterioration state that indicates deterioration that has not progressed to each deterioration state is not referred to as a non-failure state. And the usage time estimation part 12 estimates the average usage time required for a steel pipe column to reach from each assumed non-failure state to each assumed failure state based on inspection information.

  The second derivation unit 13 is a deterioration estimated by the usage time estimation unit 12 at a time t which is a variable of a functional expression representing a failure rate at each usage time of each steel pipe column derived by the first derivation unit 11. Add each average usage time for each state. Each average usage time for each deterioration state is a time required from each deemed non-failure state to each deemed failure state. As a result, the second deriving unit 13 derives a more appropriate function equation of the average failure rate curve reflecting the “deterioration state” at the time of the latest inspection.

  The future failure rate calculation unit 14 inputs “year of use” and “deterioration state” at a certain predicted time point T1 to the function formula of the average failure rate curve derived by the second deriving unit 13, Calculate the predicted failure rate of steel pipe columns according to “year” and “degradation state”.

(Inspection information data)
FIG. 2 is a diagram illustrating an example of inspection information data. The inspection information data 1 a is stored in the inspection information database 1. The inspection information database 1 is a database that records at least basic attribute information of steel pipe columns and the latest inspection results. Here, the evaluation result of the degradation state which is the inspection result is at least three states or more. The inspection information database 1 records the ground corrosion situation of the steel pipe column as an example of the inspection information data 1a.

  As shown in FIG. 2, the inspection information data 1 a includes columns of “prefecture area”, “equipment number”, “construction date”, “latest inspection date”, and “deterioration state”. “Prefectural area” is, for example, information indicating an area where a steel pipe column corresponding to the record is installed. Instead of “prefecture area”, “location information” such as latitude and longitude may be adopted. “Equipment number” is, for example, identification information for uniquely identifying a steel pipe column corresponding to the record. “Construction date” is, for example, time information indicating the date when the steel pipe column corresponding to the record is installed. “Latest inspection date” is, for example, time information indicating the date when the steel pipe column corresponding to the record was last inspected by the inspection worker. The “degraded state” is, for example, a stepwise result obtained by quantitatively and / or qualitatively evaluating a steel pipe column corresponding to the record according to a predetermined evaluation item when the inspection worker was last inspected. It is information to show.

  “Degraded state” includes, for example, states from “1” to “4”. “Deterioration state” “1” indicates a state in which the deterioration is most advanced. “Degraded state” “2” indicates a state in which the deterioration has progressed after “Degraded state” “1”. “Deterioration state” “3” indicates a state in which the deterioration has progressed after “Deterioration state” “2”. “Degraded state” “4” indicates a state in which the deterioration is least advanced or a non-degraded state. In the example shown in FIG. 2, for example, the steel pipe columns of “prefecture” “A”, “equipment number” “XX1”, “construction date” “197904 (April 1979)” “20111 (November 2011)”, “deterioration state” “1”.

(Details of processing of each part of the failure prediction device)
FIG. 3 is a diagram illustrating an example of details of processing of each unit of the failure prediction apparatus. The first deriving unit 11 of the failure predictor 10 extracts the “usage time” and “deterioration state” of the steel pipe column from the inspection information database 1 when deriving the functional equation of the average failure rate curve. As the “use time”, “year” that is a difference obtained by subtracting “construction date” from “latest check date” of each record of the check information database 1 may be adopted. Differences less than “years” may be processed according to a predetermined rounding rule such as rounding off. In addition, the unit of time of “Usage time” may be “Years” or “Months” according to “Construction date” and “Latest check date”. In addition, when a more detailed time is defined, a time unit along the detailed time may be used.

Then, the first deriving unit 11 derives a functional expression indicating the failure probability of reaching the “degraded state” “1” from the “degraded state” “2” to “4”. The first deriving unit 11 sets the data of “degraded state” “1”, which is most deteriorated, as failure data, and sets all the data in the degraded state that has not achieved failure as unfailed data (censored data). Set. Then, the first deriving unit 11 derives a function equation (Equation 2) of the average failure rate curve by Weibull analysis (cumulative hazard type). Η _avg obtained at this time is called a characteristic life, and is a state without deterioration, that is, a use time (that is, a life) in which “deterioration state” “2” to “4” reaches “deterioration state” “1”. You can think about it.

Further, the usage time estimation unit 12 estimates the usage time from “degraded state” “3” to “4” to “degraded state” “2”. First, the use time estimation unit 12 extracts “use time” and “deterioration state” of each steel pipe column from the inspection information database 1, and sets “deterioration state” “1” and “2” data as failure data. Then, the data of “degraded state” “3” and “4” is set as unfailed data. Unfailed data is censored data. Then, the usage time estimation unit 12 estimates the characteristic life (scale parameter) η ₂ by Weibull analysis (cumulative hazard type). The characteristic life η ₂ is a usage time for reaching the “degraded state” “2” from the “degraded state” “3” to “4”.

Similarly, the use time estimation unit 12 extracts “use time” and “deterioration state” of each steel pipe column from the inspection information database 1, and uses the data of “deterioration state” “1” to “3” as failure data. Set the “degraded state” “4” data as unfailed data. Unfailed data is censored data. Then, the usage time estimation unit 12 estimates the characteristic life (scale parameter) η ₃ by Weibull analysis (cumulative hazard type). The characteristic life η ₃ is a usage time from the “degraded state” “4” to the “degraded state” “3”.

The second deriving unit 13 corrects the functional equation of the average failure rate curve shown in (Equation 2) using the characteristic lifetimes η ₂ and η ₃ estimated by the usage time estimating unit 12. That is, as shown in the following (Equation 3), the second deriving unit 13 derives a function formula of a future failure rate curve reflecting not only the use time t but also the “deterioration state” at the time of the latest inspection. . In other words, for the non-failed data of “2” to “4” in the “deterioration state” at the time of the latest inspection, further deterioration from “deterioration state” “4” occurs in the usage time t in the function formula of the average failure rate curve. Is added to the life characteristic η ₂ or η ₃ which has been advanced.

  In other words, when predicting the failure rate at the future time point after the present, the usage time in the function formula of the average failure rate curve for unfailed data with a deterioration state other than “no deterioration” at the time of the latest inspection Add to t the average time to deteriorate from the no deterioration state to the “deterioration state” at the latest inspection. Thereby, a more reasonable failure rate reflecting the “deterioration state” at the time of the latest inspection can be predicted.

  Then, the future failure rate calculation unit 14 inputs the actual years of use of the steel pipe column at the predicted time T1 after the current time and the “deterioration state” at the latest inspection, and calculates the failure rate at the predicted time T1 after the current time. And the future failure rate calculation part 14 orders each steel pipe column in descending order of the failure rate of the prediction time T1 after the present time predicted about each steel pipe column, for example, controls the output apparatus 2 grade | etc., And outputs it.

(Weibull analysis (cumulative hazard type))
The Weibull analysis (cumulative hazard type) will be described below. The cumulative hazard function H (t) is obtained by integrating the failure rate of the Weibull distribution shown in the following (Equation 4) in the time interval [0, t] (t> 0) as shown in the following (Equation 5). Is.

  Then, taking the natural logarithm of both sides of the above (Formula 5), the following (Formula 6) is obtained.

  Then, with respect to “ln (H (t))” on the left side of (Equation 6) and “ln (t)” on the right side, “ln (t)” is represented on the horizontal axis and by the cumulative hazard analysis based on random truncation. Weibull plotting is performed on a graph having “ln (H (t))” as a vertical axis. Here, x = ln (t) is set and y = ln (H (t)) is set. The slope of the straight line when the Weibull plot is linearly approximated by the method of least squares or the like is the estimated value of the shape parameter m in the above (Expression 4). Further, a scale parameter (or characteristic life) η is estimated from ln (t) when ln (H (t)) = 0 with respect to the estimated shape parameter m in a straight line obtained by linear approximation. . For cumulative hazard analysis based on random truncation, we use existing methods such as “Kenji Tanaka,“ Introductory Reliability Engineer Learns for the First Time ”, Nikka Giren Publisher, December 25, 2008, pp.102-105”. Can be used.

(Example of Weibull plot)
An example of the Weibull plot will be described with reference to FIGS. FIG. 4 is a diagram illustrating an example of a Weibull plot when the degradation state 1 is a failure. In FIG. 4, “degraded state” “1” is regarded as a failure, “degraded states” “2” to “4” are regarded as no failure, cumulative hazard analysis based on the above-mentioned random censoring is performed, and a Weibull plot is performed. Show the case. As shown in FIG. 4, the Weibull plot can be linearly approximated as “y = 5.307x−23.29”. Here, “y = ln (H (t))” and “x = ln (t)”. Therefore, the shape parameter m _avg and the scale parameter (or characteristic life) η _avg of the functional equation of the average failure rate curve shown in (Expression 2) can be estimated by the Weibull plot shown in FIG. m _avg = 5.307 and η _avg = 80.52. The function equation of the average failure rate curve based on the Weibull distribution when “degraded state” “1” is regarded as a failure and “degraded states” “2” to “4” is regarded as no failure is as follows: It is as (Formula 7).

FIG. 5 is a diagram illustrating an example of a Weibull plot when the degradation states 1 and 2 are failures. FIG. 5 considers “degraded states” “1” and “2” as failures, “degraded states” “3” and “4” as no failures, performs cumulative hazard analysis based on the above-mentioned random truncation, and performs Weibull The case where plotting was performed is shown. As shown in FIG. 5, the Weibull plot can be linearly approximated as “y = 4.679x−19.63”. Here, “y = ln (H (t))” and “x = ln (t)”. Therefore, according to the Weibull plot shown in FIG. 5, the shape parameter m (here, m ₂ ) and the scale parameter (or characteristic life) η (here η) of the average failure rate curve shown in (Equation 4) above. ₂ ). m ₂ = 4.679 and η ₂ = 66.37. It should be noted that a functional expression of an average failure rate curve based on the Weibull distribution when “degraded state” “1” and “2” are regarded as failures and “degraded state” “3” and “4” are regarded as no failures. Is as in (Equation 8) below.

FIG. 6 is a diagram illustrating an example of a Weibull plot when the degradation states 1 to 3 are failures. In FIG. 6, “degraded state” “1” to “3” is regarded as a failure, “degraded state” “4” is regarded as no failure, cumulative hazard analysis based on the above-mentioned random censoring is performed, and a Weibull plot is performed. Show the case. As shown in FIG. 6, the Weibull plot can be linearly approximated as “y = 3.601x-14.06”. Here, “y = ln (H (t))” and “x = ln (t)”. Therefore, the Weibull plot shown in FIG. 6, the average failure rate curve of function expression of shape parameter m as shown in the above equation (4) (in this case is denoted by m ₃₎ and the scale parameter (or characteristic life) eta (here eta ₃ ) can be estimated. m ₃ = 3.601 and η ₃ = 49.62. In addition, when the “degraded state” “1” to “3” is regarded as a failure and the “degraded state” “4” is regarded as no failure, a functional equation of an average failure rate curve based on the Weibull distribution is as follows: It is as (Formula 9).

  FIG. 7 shows combinations of each shape parameter m and scale parameter (or characteristic life) η obtained by Weibull plot as described above. “Case 1” in FIG. 7 is a case where “degraded state” “1” is regarded as a failure and “degraded states” “2” to “4” are regarded as no failure. “Case 2” in FIG. 7 is a case where “degraded states” “1” and “2” are regarded as failures, and “degraded states” “3” and “4” are regarded as no failures. Further, “Case 3” in FIG. 7 is a case where “degraded state” “1” to “3” is regarded as a failure, and “degraded state” “4” is regarded as no failure.

(Failure prediction processing)
FIG. 8 is a flowchart showing failure prediction processing. The first deriving unit 11 of the failure prediction apparatus 10 extracts information on the usage time and the deterioration state from the inspection information database 1, and obtains the shape parameter m _avg and the scale parameter η _avg by Weibull analysis (cumulative hazard type). Then, the first deriving unit 11 derives a functional expression of an average failure rate curve based on the shape parameter m _avg and the scale parameter η _avg (step S11).

  Next, the usage time estimation unit 12 of the failure prediction apparatus 10 extracts the usage time and deterioration state information from the inspection information database 1, and estimates the usage time to reach each deterioration state by Weibull analysis (cumulative hazard type). (Step S12).

  Next, the second derivation unit 13 of the failure prediction apparatus 10 uses the function formula of the average failure rate curve and the usage time to reach each deterioration state, and can reflect the deterioration state at the latest inspection from the present to the future in the future. A function formula of the failure rate curve is derived (step S13).

  Next, the future failure rate calculation unit 14 of the failure prediction apparatus 10 determines the estimated time T1 after the current time based on the actual years of use at the predicted time T1 after the current time and the deterioration state at the latest inspection for each steel pipe column. A failure rate is calculated (step S14).

(Effect by embodiment)
According to the embodiment, by correcting the function formula of the average failure rate curve using the use time until each deterioration state is reached, there is a difference in deterioration state at the time of the latest inspection that cannot be expressed only by the actual years of use. It is possible to predict a more appropriate equipment failure rate that is taken into consideration. As a result, the accuracy of failure (deterioration) prediction can be improved, and efficient maintenance, its planning, and decision support can be performed.

(Modification of the embodiment)
In the embodiment, a functional expression of a future failure rate curve based on a Weibull distribution that can be expressed by a power function at time t is derived. However, not only the power function at time t, but any function expression that can express the future failure rate, such as an exponential function at time t, a logarithmic function at time t, or a bathtub curve, may be derived. And the constant added to the time t according to the deterioration state at the time of the latest inspection is calculated with respect to any function formula which can express a future failure rate. Then, the failure rate at the predicted time T1 after the current time may be calculated based on the actual years of use at the predicted time T1 after the current time and the deterioration state at the time of the latest inspection.

  In the embodiment, the rank classification of the deterioration state at the time of the latest inspection is three stages, but may be four stages or more. By further classifying the degradation state into four or more stages, it is possible to predict a more appropriate equipment failure rate in consideration of the difference in the degradation state at the latest inspection.

(Application example of embodiment)
Below, the example of application of embodiment is shown. FIG. 9A is a diagram showing a ranking of failure rates calculated by the conventional technique. FIG. 9B is a diagram showing the rank of failure rates calculated by the technique of the embodiment. In FIG. 9A and FIG. 9B, it is assumed that the number of data of inspection results of steel pipe columns in a certain prefecture area is, for example, 16,239. Then, each “degraded state” of the inspection result is divided into four ranks “1” to “4”, and “degraded state” “1” is the state in which the deterioration is most advanced and is subject to renewal. It is assumed that the degree of deterioration is reduced in the order of “deterioration state” “2”, “3”, “4”, and “deterioration state” “4” is regarded as no deterioration.

  The Weibull plot (case 1) in the case where the data of “degraded state” “1” is treated as a failure and the data of “degraded state” “2”, “3”, “4” is not failed is as shown in FIG. It is. The Weibull plot (Case 2) in the case where the data of “degraded state” “1” and “2” is treated as a failure and the data of “degraded state” “3” and “4” is not failed is shown in FIG. As shown in Also, the Weibull plot (Case 3) when the data of “degraded state” “1” to “3” is treated as a failure and the data of “degraded state” “4” is not failed is as shown in FIG. is there. The combinations of the shape parameter m and the scale parameter (or characteristic life) η in each case obtained by the Weibull plot are as shown in FIG.

  For example, about 200 pieces of data are randomly extracted from all 16,239 data, and the current failure rate is calculated and the facility inspection is prioritized by the method of the existing technology and the method of the embodiment. When the latest inspection result is “degraded state” “4”, the failure rate is obtained by the following (formula 10) based on the above (formula 2). When the latest inspection result is “degraded state” “2”, the failure rate is obtained from the following (formula 11) based on the above (formula 3). Similarly, when the latest inspection result is “degraded state” “3”, the failure rate is obtained by the following (formula 12).

  In the result by the existing method shown in FIG. 9A, since the failure rate depends only on the service life, steel pipe columns with a longer service life are ranked higher. When there are many steel pipe columns having the same service life, there are many steel pipe columns having the same failure rate. Therefore, when maintenance resources (human resources, equipment resources, time, funds, etc.) are limited for many steel pipe columns, and all of them cannot be checked at once, the existing method sets the priority of the steel pipe columns to be maintained. It is difficult to put on.

  However, as shown in FIG. 9B, if the method of the embodiment is applied, it can be confirmed that there are many cases where a ranking can be given to steel pipe columns that cannot be prioritized by the existing method. In the result according to the embodiment shown in FIG. 9B, since the failure rate depends on the service life and the deterioration state, even a steel pipe column having a long service life may not be ranked higher depending on the deterioration state. Further, even steel pipe columns with a short service life may be ranked higher depending on the deterioration state. That is, even when there are many steel pipe columns having the same service life, there are many cases where different failure rates are obtained by taking the deterioration state into consideration. Therefore, even if maintenance resources are limited for many steel pipe columns and not all of them can be inspected at the same time, in the embodiment, the priority order of the steel pipe columns to be maintained can be set.

(About device configuration of failure prediction device)
Each component of the failure prediction apparatus shown in FIG. 1 is functionally conceptual and does not necessarily need to be physically configured as illustrated. That is, the specific form of distribution and integration of the functions of the failure prediction apparatus 10 is not limited to the illustrated one, and all or part of the functions can be functionally or physically in arbitrary units according to various loads or usage conditions. Can be distributed or integrated.

  In addition, all or any part of each process performed in the failure prediction apparatus 10 may be realized by a CPU (Central Processing Unit) and a program that is analyzed and executed by the CPU. In addition, each process performed in the failure prediction apparatus 10 may be realized as hardware by wired logic.

  In addition, among the processes described in the embodiment, all or a part of the processes described as being automatically performed can be manually performed. Alternatively, among the processes described in the embodiments, all or a part of the processes described as being performed manually can be automatically performed by a known method. In addition, the above-described and illustrated processing procedures, control procedures, specific names, and information including various data and parameters can be changed as appropriate unless otherwise specified.

(About the program)
FIG. 10 is a diagram illustrating an example of a computer in which the failure prediction apparatus is realized by executing each program. The computer 1000 includes a memory 1010 and a CPU 1020, for example. The computer 1000 also includes a hard disk drive interface 1030, a disk drive interface 1040, a serial port interface 1050, a video adapter 1060, and a network interface 1070. In the computer 1000, these units are connected by a bus 1080.

  The memory 1010 includes a ROM (Read Only Memory) 1011 and a RAM (Random Access Memory) 1012. The ROM 1011 stores a boot program such as BIOS (Basic Input Output System). The hard disk drive interface 1030 is connected to the hard disk drive 1031. The disk drive interface 1040 is connected to the disk drive 1041. For example, a removable storage medium such as a magnetic disk or an optical disk is inserted into the disk drive 1041. The serial port interface 1050 is connected to a mouse 1051 and a keyboard 1052, for example. The video adapter 1060 is connected to the display 1061, for example.

  The hard disk drive 1031 stores, for example, an OS 1091, an application program 1092, a program module 1093, and program data 1094. That is, a program defined by each process of the failure prediction apparatus 10 is stored in, for example, the hard disk drive 1031 as a program module 1093 in which a command executed by the computer 1000 is described. For example, a program module 1093 for executing information processing similar to the functional configuration in the failure prediction apparatus 10 is stored in the hard disk drive 1031.

  The setting data used in the processing in the above-described embodiment is stored as program data 1094, for example, in the memory 1010 or the hard disk drive 1031. Then, the CPU 1020 reads the program module 1093 and the program data 1094 stored in the memory 1010 and the hard disk drive 1031 to the RAM 1012 as necessary, and executes them.

  Note that the program module 1093 and the program data 1094 are not limited to being stored in the hard disk drive 1031, but may be stored in, for example, a removable storage medium and read out by the CPU 1020 via the disk drive 1041 or the like. Alternatively, the program module 1093 and the program data 1094 may be stored in another computer connected via a network (LAN (Local Area Network), WAN (Wide Area Network), etc.). The program module 1093 and the program data 1094 may be read by the CPU 1020 via the network interface 1070.

  The above embodiments and modifications thereof are included in the invention disclosed in the claims and equivalents thereof as well as included in the technology disclosed in the present application.

1 Inspection Information Database 2 Output Device 10 Failure Prediction Device 11 First Deriving Unit 12 Usage Time Estimating Unit 13 Second Deriving Unit 14 Future Failure Rate Calculation Unit 1000 Computer 1010 Memory 1020 CPU

Claims (7)

Deriving a function formula for the average failure rate curve that predicts the average failure rate for a certain usage time of the managed object from data that accumulates the inspection results of each managed object that is divided into three or more deterioration levels in stages. A first derivation unit that
An estimation unit for estimating each average usage time until each managed object reaches each degradation state;
By correcting the function formula of the average failure rate curve derived by the first deriving unit with each average usage time estimated by the estimation unit, the usage time of each managed object corresponding to each degradation state A second derivation unit for deriving a functional expression of a failure rate curve for predicting a failure rate at
Using a functional expression of a failure rate curve derived by the second deriving unit, and a calculation unit that calculates each failure rate at which each managed object corresponding to each degradation state fails in a certain usage time. Feature failure prediction device. In the data, the actual usage time of each managed object is associated with the deteriorated state of each managed object in the first to nth deterioration states (where n is a natural number satisfying n ≧ 3). And
The first deriving unit sets the first deterioration state as a failure state, sets the second to nth deterioration states as a non-failure state, and the management target object corresponding to the non-failure state in the data is: Deriving a functional equation of the average failure rate curve that represents the failure rate to meet the failure state as a function of usage time,
The estimation unit sets the first to kth deterioration states (where k is a natural number satisfying 1 <k <n) as a failure state, and sets the (k + 1) th to nth deterioration states as a non-failure state. Set, and estimate the k-th average usage time until the management target object corresponding to the non-failure state corresponds to the failure state in the data,
The second deriving unit adds the k-th average usage time to the usage time in the functional equation of the average failure rate curve, and the management target corresponding to the k-th degradation state in the data is the first The failure prediction apparatus according to claim 1, wherein a function expression of the failure rate curve that represents the failure rate that corresponds to a deterioration state of the device as a function of usage time is derived. The first derivation unit derives a functional expression of the average failure rate curve based on Weibull analysis,
The failure prediction apparatus according to claim 1, wherein the estimation unit estimates the average usage time based on Weibull analysis. The first derivation unit of the failure prediction apparatus calculates the average failure rate during a certain usage time of the managed object from the data obtained by accumulating the inspection results of the managed objects that are classified in stages into three or more deterioration states. Deriving a function equation for the predicted mean failure rate curve,
The estimation unit of the failure prediction apparatus estimates each average usage time until each managed object corresponds to each deterioration state,
The second derivation unit of the failure prediction apparatus corrects the function expression of the average failure rate curve derived by the first derivation unit with each average usage time estimated by the estimation unit, thereby reducing each deterioration. Deriving a functional equation of failure rate curve that predicts the failure rate of each managed object corresponding to the state at a certain usage time,
The calculation unit of the failure prediction device uses the functional expression of the failure rate curve derived by the second deriving unit to calculate each failure rate at which each managed object corresponding to each degradation state fails in a certain usage time. A failure prediction method characterized by including calculating. In the data, the actual usage time of each managed object is associated with the deteriorated state of each managed object in the first to nth deterioration states (where n is a natural number satisfying n ≧ 3). And
The first derivation unit sets the first deterioration state as a failure state, sets the second to nth deterioration states as a non-failure state, and the management target object corresponding to the non-failure state in the data includes: Deriving a functional equation of the average failure rate curve that represents the failure rate to meet the failure state as a function of usage time,
The estimation unit sets the first to kth deterioration states (where k is a natural number satisfying 1 <k <n) as a failure state, and sets the (k + 1) to nth deterioration states as a non-failure state. Set, and estimate the k-th average usage time until the management target object corresponding to the non-failure state corresponds to the failure state in the data,
The second deriving unit adds the k-th average usage time to the usage time in the functional expression of the average failure rate curve, and the management target corresponding to the k-th degradation state in the data is the first The failure prediction method according to claim 4, further comprising: deriving a functional expression of the failure rate curve that represents the failure rate that corresponds to the deterioration state of the device as a function of usage time. The first derivation unit derives a functional expression of the mean failure rate curve based on Weibull analysis;
The failure prediction method according to claim 4, wherein the estimation unit includes estimating the average usage time based on Weibull analysis.   A failure prediction program for causing a computer to function as the failure prediction apparatus according to claim 1.

Images