JP2013114636A - Maintenance support system, maintenance support method and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide precise and sufficient information as maintenance support information for planing maintenance of equipment.SOLUTION: An estimated remaining quantity rate 31, the actual number of pieces of equipment installed 32 and the actual number of repaired pieces of equipment 33 are data acquired by acquisition processing 61a, 61b and 61c, respectively. By estimated number of remaining pieces of equipment calculation processing 62, an estimated number of remaining pieces of equipment 41 is calculated for each lapse period on the basis of the estimated remaining quantity rate 31 and the actual number of pieces of equipment installed 32. By estimated hazard function calculation processing 63, an estimated hazard function 42 is calculated on the basis of the estimated number of remaining pieces of equipment 41 and the actual number of repaired pieces of equipment 33. By estimated accumulated repair rate calculation processing 64, an estimated accumulated repair rate 43 is calculated on the basis of the estimated hazard function 42. By prediction processing 65a, a predicted accumulated repair rate 51 is calculated by applying a prediction model 22 for the estimated accumulated repair rate 43 by using data within a range of a learning period 23.

Description

  The present invention relates to a maintenance support system that supports maintenance of equipment used mainly in general households.

  In maintenance work of various devices, it is required to respond quickly when a failure occurs and to make an appropriate maintenance plan so that a serious failure does not occur. The specific contents of the maintenance plan cover a variety of topics, such as discontinuation of equipment sales, parts storage age, extended warranty period, and design changes.

  Patent Document 1 discloses a quality control method for gas equipment used in general households. In the quality control method described in Patent Document 1, first, for all owned devices registered in the customer-owned device data storage unit, the number of devices according to the length of the unit period actually used is classified by device type. Count to calculate the number of units by device type and period of use. Next, based on the component usage record data, the number of used parts for each used part is derived by equipment type and used period, and this is divided by the number of used parts by used equipment type. Of the owned devices, the component usage rate, which is a value that defines the ratio of the number of components that are required to be used within the unit period for each component type and the usage period for each device type, is calculated as a quality control index.

Patent Document 2 discloses a maintenance support program that supports estimation work of device replacement time and maintenance time. The maintenance support program of Patent Document 2 is a program for causing a computer to realize the following functions.
(1) A set of devices shipped to a predetermined shipping destination based on shipping information stored in a shipping information database and associated with identification information of a device to be maintained and customer identification information indicating a shipping destination of the device. A population for obtaining the number Ni of a population including a device that has failed at a certain shipping destination by the time of reliability analysis and a device that has not failed at the predetermined shipping destination by the time of reliability analysis Arithmetic function.
(2) Reliability analysis within the population based on equipment identification information stored in the equipment maintenance information database, customer identification information indicating customers using the equipment, and equipment maintenance information associated with equipment failure occurrence time information Cumulative hazard value (H (t) = number of failed devices / Ni having failed until the time of reliability analysis at a predetermined shipping destination) is accumulated and accumulated. Based on the environmental information that is described in the hazard value and usage environment database and associates customer identification information with the usage environment of the device, cumulative hazard function approximation, reliability function calculation, unreliability function calculation, failure probability Perform reliability analysis including at least one of density function calculation and Weibull function calculation while performing weight correction according to the usage environment, and reliability analysis for a given shipping destination Analysis function of recording reliability analysis information indicating a result in reliability analysis database.

JP 2010-231376 A Japanese Patent No. 4309803

  By the way, in Patent Document 1, based on the “erasure date” (the date when the maintenance company recognizes that the device has been removed) registered in the customer-owned device data storage unit, “the length of the actually used unit period” "The number of devices according to the size" is demanded. In that case, if the customer disposes of the device or leaves it without using it without contact for repair after sale / installation, it will be counted as a device that is operating normally. . Therefore, the component usage rate calculated in Patent Document 1 cannot be said to be a quality control index based on accurate grasp of actual conditions. Furthermore, in patent document 1, it is not sufficient as information for making an appropriate maintenance plan, only to the point where the component usage rate is calculated as a quality control index.

  In Patent Document 2, the number of operating units at the time of reliability analysis is calculated by subtracting failure information from shipping information to the time of reliability analysis. Then, as mentioned above, if the customer disposes of the device or leaves it unused without using it for repair after sales / installation, the device is operating normally. It will be counted. Therefore, the reliability analysis information calculated in Patent Document 2 cannot be said to be a quality control index based on an accurate actual grasp. In addition, in Patent Document 2, “failures with different properties” in which the failure location is different or the elapsed time is different even if the failure location is the same is not considered separately. Not enough information.

  The present invention has been made in view of the above-described problems, and its purpose is to provide maintenance support capable of providing accurate and sufficient information as maintenance support information for making a maintenance plan for equipment. It is to provide a system and the like.

  In order to achieve the above-described object, a first invention is a maintenance support system configured by a computer and supporting a maintenance operation of a device, wherein the device is actually installed for each elapsed period from the installation time of the device. Storage means for storing an estimated remaining rate that is an estimated value of the operating ratio, acquisition means for acquiring the actual number of installed devices for each elapsed period, and acquiring the number of actual repairs for the device for each occurrence time And an estimated remaining number calculating means for calculating an estimated remaining number that is an estimated value of the number of devices actually remaining for each elapsed period based on the estimated remaining rate and the actual number of installed units; Based on the number of repairs and the estimated remaining number of units, estimated hazard function calculating means for calculating an estimated hazard function that is an estimated value of the hazard function of the device, and based on the estimated hazard function A maintenance support system characterized by comprising an estimation cumulative repair rate calculating means for calculating an estimate of the cumulative repair rate is estimated cumulative repair rate of the device.

  According to the first invention, the maintenance support system can provide accurate information as maintenance support information for making a maintenance plan for the equipment. In particular, if the estimated remaining rate stored in the storage means reflects, for example, the results of a periodic security survey that periodically visits general households where equipment is installed, the maintenance support system is Maintenance support information can be provided based on accurate grasp of the actual situation.

  The first invention applies a predetermined prediction model to the estimated cumulative repair rate using a reception unit that receives selection of a learning period and data within the range of the learning period, and the future accumulation of the device It is desirable to further include prediction means for performing a prediction process for calculating a predicted cumulative repair rate that is a predicted value of the repair rate. Thus, the user can dynamically change the learning period and obtain maintenance support information (for example, a predicted cumulative repair rate). Therefore, the user can select an appropriate learning period according to past cases (for example, cases such as breakdowns and repairs) and the like, and can obtain accurate maintenance support information.

  In the prediction process according to the first aspect of the invention, a predicted remaining number that is a predicted value of a future remaining number of the devices, a predicted hazard function that is a predicted value of a future hazard function of the device, and a future repair of the device It is desirable to calculate a predicted repair rate that is a predicted value of the rate and a predicted repair number that is a predicted value of the future number of repairs of the device. Thereby, the user can obtain various information as the maintenance support information.

  It is desirable that the storage means in the first invention further comprises display means for storing maintenance execution information relating to the device at each occurrence time point and displaying the maintenance execution information. Accordingly, the user can select an appropriate learning period while confirming the maintenance execution information.

  The receiving means in the first invention preferably further has a function of limiting a selectable range of the learning period based on the maintenance execution information. Accordingly, the maintenance support system can make it impossible to select a period that should not be included in the learning period from past cases. For example, when parts replacement is performed for all sold devices and the tendency of occurrence of failure or repair has changed significantly, the maintenance support system cannot select a period before the part replacement time. Thus, accurate maintenance support information can be calculated.

  The receiving means in the first invention preferably further has a function of automatically determining the optimum learning period based on the maintenance execution information. Thereby, the maintenance support system can save the user's trouble and can provide accurate maintenance support information to a user who is not familiar with the system. For example, when parts replacement is performed for all sold devices and the tendency of occurrence of failure or repair has changed significantly, the maintenance support system should determine the period after the part replacement time as the learning period Thus, accurate maintenance support information can be calculated.

  In the first aspect of the present invention, it is preferable that the accepting unit accepts selection of a plurality of learning periods, and the predicting unit performs a plurality of prediction processes for each of the plurality of learning periods and compares and outputs a prediction result. . Thereby, the user can compare the prediction results of a plurality of learning periods. Therefore, the user can find out that maintenance work that causes a trend change in the past has been performed.

  In the first aspect of the present invention, the accepting unit accepts selection of a past prediction period, the prediction unit performs the prediction process for the past prediction period, and predicts the past prediction period calculated by the prediction process. It is desirable to further include an outlier calculation means for comparing the value with an actual value in the same period as the past prediction period and calculating an outlier from the actual value. As a result, the user can confirm the degree of deviation from the actual value. Therefore, the user can verify the validity of the prediction model used in the prediction process and the learning period.

  And determining means for determining whether or not the degree of detachment calculated by the degree of detachment calculating means in the first invention is greater than or equal to a predetermined threshold. The degree of detachment is determined to be greater than or equal to a predetermined threshold by the determining means. When it is determined, the receiving unit receives re-selection of the learning period, and when the determining unit determines that the degree of detachment is less than a predetermined threshold, the predicting unit It is desirable to calculate This eliminates the need for the user himself to change the learning period or the like and further determine whether or not the maintenance support system should execute the prediction process. Furthermore, the user can confirm the confidence interval of the prediction result. Therefore, the maintenance support system can give the user a sense of satisfaction with the maintenance support information.

  The prediction means in the first invention preferably further has a function of determining the learning period one by one from all possible periods, repeating the prediction process, and determining the optimum learning period. As a result, the maintenance support system can perform prediction processing for all possible periods, and can automatically determine the optimal learning period from the prediction results.

  The storage means in the first invention further stores repair information including a failure phenomenon of the device, a cause of the failure, a repair location and a repair method, and the reception means further includes a data range used for the prediction process. As a narrowing-down condition, selection of one or a plurality of values of the failure phenomenon, the cause of the failure, the repair location, and the repair method is accepted, and the predicting unit is in accordance with the narrowing-down condition received by the receiving unit. It is desirable to perform the prediction process using data to be narrowed down. As a result, the user can select one or more values of the failure phenomenon, the failure cause, the repair location, and the repair method to narrow down the data range and obtain the prediction support information. As a result, the user can obtain various prediction support information according to the purpose of the maintenance work.

  In the first invention, it is preferable that the accepting unit accepts selection of the prediction model, and the predicting unit performs the prediction process by applying the prediction model received by the accepting unit. Thereby, the user can select the prediction model in addition to the learning period, and can obtain the prediction support information.

  A second invention is a maintenance support method executed by a computer and supporting a maintenance operation of the device, wherein the computer is actually operating for each elapsed period from the installation time of the device. A storage step of storing an estimated remaining rate that is an estimated value of a ratio; an acquisition step of acquiring the actual number of installed devices for each elapsed period; and acquiring an actual number of repairs of the device for each occurrence time; and the estimation Based on the remaining rate and the actual installed number, an estimated remaining number calculating step for calculating an estimated remaining number that is an estimated value of the number of devices actually remaining for each elapsed period; Based on the estimated remaining number, an estimated hazard function calculating step for calculating an estimated hazard function that is an estimated value of the hazard function of the device, and on the basis of the estimated hazard function, The estimated cumulative repair rate calculating step of calculating an estimated cumulative repair rate is an estimate of the cumulative repair rate of the serial device, a maintenance supporting method which comprises a. According to the second invention, accurate information can be provided as maintenance support information for making a maintenance plan for equipment.

  The second invention applies a predetermined prediction model to the estimated cumulative repair rate using a reception step for accepting selection of a learning period and data within the range of the learning period, and the future accumulation of the device It is desirable to further include a prediction step for performing a prediction process for calculating a predicted cumulative repair rate that is a predicted value of the repair rate. As a result, the user can dynamically change the learning period and obtain maintenance support information. Therefore, the user can select an appropriate learning period according to past cases and the like, and can obtain accurate maintenance support information.

  A third invention is a program for causing a computer to execute a maintenance support method for supporting a maintenance operation of a device, and the device is actually operated for each elapsed period from the installation time of the device. A storage step of storing an estimated remaining rate that is an estimated value of a ratio; an acquisition step of acquiring the actual number of installed devices for each elapsed period; and acquiring an actual number of repairs of the device for each occurrence time; and the estimation Based on the remaining rate and the actual installed number, an estimated remaining number calculating step for calculating an estimated remaining number that is an estimated value of the number of devices actually remaining for each elapsed period; An estimated hazard function calculating step for calculating an estimated hazard function that is an estimated value of the hazard function of the device based on the estimated remaining number of units, and based on the estimated hazard function Is a program for executing the processing including an estimation cumulative repair rate calculating step of calculating the estimated value and is estimated cumulative repair rate of accumulation repair rate of the device to the computer. By installing the third invention in a computer, the maintenance support system of the first invention can be obtained. As a result, accurate information can be provided as maintenance support information for making a maintenance plan for equipment.

  A third aspect of the invention is to apply a predetermined prediction model to the estimated cumulative repair rate using a reception step of accepting selection of a learning period and data within the range of the learning period, and It is desirable to cause the computer to execute a process that further includes a prediction step of performing a prediction process of calculating a predicted cumulative repair rate that is a predicted value of the repair rate. As a result, the user can dynamically change the learning period and obtain maintenance support information. Therefore, the user can select an appropriate learning period according to past cases and the like, and can obtain accurate maintenance support information.

  According to the present invention, it is possible to provide a maintenance support system that can provide accurate and sufficient information as maintenance support information for making a maintenance plan for equipment.

The figure which shows the outline of the maintenance support system 1 Hardware configuration diagram of terminal 2 (server 3) Data flow showing the data flow of the maintenance support system 1 An example of device installation information 70 An example of estimated remaining rate information 80 An example of repair information 90 An example of maintenance execution information 100 The flowchart which shows the flow of a process of the maintenance support system 1 Display example of maintenance execution information 100 and selectable range of learning period 23 The flowchart which shows the process of the modification 2. The flowchart which shows the process of the modification 3. Output example of graph data 56 The flowchart which shows the process of the modification 4. The flowchart which shows the process of the modification 5.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. Below, a gas apparatus is mentioned as an example and mainly demonstrated as an apparatus mainly used in a general household. However, the scope of application of the present invention is not limited to gas equipment. If the present invention is applied to a device having the above-described problems, the same effect can be obtained.

FIG. 1 is a diagram showing an outline of the maintenance support system 1. As shown in FIG. 1, in the maintenance support system 1, for example, a terminal 2 and a server 3 are connected via a network 6. The network 6 is, for example, the Internet or a LAN (Local Area Network). The terminal 2 is, for example, a PC (Personal
Any device that can connect to the network 6 and perform data communication (HTTP communication, TCP / IP communication, etc.), such as a computer (hereinafter referred to as “computer”) or a mobile terminal (mobile phone, smartphone, tablet terminal, etc.) But it ’s okay. Similarly to the terminal 2, the server 3 may be connected to the network 6 and can perform data communication, but is preferably a high-performance server computer.

  A maintenance support program 4 that is an embodiment of the present invention is installed in the terminal 2. The server 3 is constructed with a database (hereinafter “DB”) 5 for storing various data used in the embodiment of the present invention. In the embodiment of the present invention, the terminal 2 functions as various means according to the maintenance support program 4 and presents maintenance support information necessary for making a maintenance plan for the device to the user. The terminal 2 transmits a data request command to the server 3 as necessary. The server 3 searches the DB 5 for the data request and transmits the requested data to the terminal 2.

  The configuration of the maintenance support system 1 is not limited to the example illustrated in FIG. For example, the maintenance support system 1 may be configured with only the terminal 2. That is, the terminal 2 may include the DB 5.

  The maintenance support program 4 may be installed in the server 3. That is, the server 3 may function as various means according to the maintenance support program 4 and present maintenance support information to the user. In this case, the terminal 2 serves as an interface with the user. That is, the terminal 2 transmits data input from the user to the server 3 and outputs (displays, prints, etc.) data received from the server 3.

  Further, instead of DB5, data may be stored as a simple file. The data stored in the DB 5 may be acquired from an external server.

  FIG. 2 is a hardware configuration diagram of a computer that realizes the terminal 2 (server 3). Note that the hardware configuration in FIG. 2 is an example, and various configurations can be adopted depending on the application and purpose.

  A computer that realizes the terminal 2 (server 3) includes a control unit 11, a storage unit 12, a media input / output unit 13, a communication control unit 14, an input unit 15, a display unit 16, a peripheral device I / F unit 17, and the like. 18 is connected.

  The control unit 11 includes a CPU (Central Processing Unit), a ROM (Read Only Memory), a RAM (Random Access Memory), and the like. The CPU calls a program stored in the storage unit 12, ROM, recording medium, etc. to a work memory area on the RAM and executes it, drives and controls each device connected via the bus 18, and will be described later. Realize processing. The ROM is a non-volatile memory and permanently holds a computer boot program, a program such as BIOS, data, and the like. The RAM is a volatile memory, and temporarily stores programs, data, and the like loaded from the storage unit 12, ROM, recording medium, and the like, and includes a work area used by the control unit 11 for performing various processes.

  The storage unit 12 is, for example, an HDD (Hard Disk Drive), and stores a program executed by the control unit 11, data necessary for program execution, an OS (Operating System), and the like. As for the program, a control program corresponding to the OS and an application program for causing a computer to execute processing to be described later are stored. Each of these program codes is read by the control unit 11 as necessary, transferred to the RAM, read by the CPU, and executed as various means.

  The media input / output unit 13 (drive device) inputs / outputs data, for example, media such as a CD drive (-ROM, -R, -RW, etc.), DVD drive (-ROM, -R, -RW, etc.) Has input / output devices. The communication control unit 14 has a communication control device, a communication port, and the like, is a communication interface that mediates communication between the computer and the network 6, and controls communication with other computers via the network 6. The network 6 may be wired or wireless.

  The input unit 15 inputs data and includes, for example, a keyboard, a pointing device such as a mouse, and an input device such as a numeric keypad. An operation instruction, an operation instruction, data input, and the like can be performed on the computer via the input unit 15. The display unit 16 includes a liquid crystal panel, a display device such as an organic EL, and a logic circuit or the like (video adapter or the like) for realizing a video function of a computer in cooperation with the display device. The input unit 15 and the display unit 16 may be integrated like a touch panel display.

  The peripheral device I / F (interface) unit 17 is a port for connecting a peripheral device to the computer, and the computer transmits and receives data to and from the peripheral device via the peripheral device I / F unit 17. The peripheral device I / F unit 17 is configured by a USB or the like, and usually includes a plurality of peripheral devices I / F. The connection form with the peripheral device may be wired or wireless. The bus 18 is a path that mediates transmission / reception of control signals, data signals, and the like between the devices.

  FIG. 3 is a data flow showing a data flow of the maintenance support system 1. FIG. 3 describes the flow of data, and specific examples of data and processing contents will be described later.

  In the following, the period in which the actual value is registered in DB5 is “past”, the period in which the actual value is not registered in DB5 is “future”, and the boundary between “past” and “future” is “present” . That is, the time axis in the embodiment of the present invention is not based on the actual time but based on whether or not the actual value is registered in the DB 5.

  The prediction target (device / part / repair location, etc.) 21, the prediction model 22, and the learning period 23 are data that the user inputs via the input unit 15 of the terminal 2 in order to obtain desired maintenance support information.

  The prediction target 21 includes a device, a part, a repair location, and the like to be predicted. The user inputs at least a device to be predicted (specified by a device code such as a model number) as the prediction target 21.

  The prediction model 22 is a statistical model used by the terminal 2 in the prediction process 65a. Examples of the statistical model include a Weibull model using a Weibull distribution, a linear model using a least square method, and the like.

  The learning period 23 is a range of data in which the terminal 2 applies the prediction model 22 in the prediction process 65a. The learning period 23 is defined as an elapsed period from the installation time of the device (hereinafter, simply referred to as “elapsed period”). The elapsed period may be yearly, such as one year, two years, and so on, or may be monthly, such as one month, two months, and so on.

  The estimated remaining rate 31, the actual installed number 32, and the actual repair number 33 are data acquired by the acquisition processes 61a, 61b, and 61c, respectively.

  The estimated remaining rate 31 is an estimated value of the rate at which the device is actually operating, which is registered in the DB 5 for each elapsed period from the installation time of the device. The acquisition process 61 a searches the DB 5 using the prediction target 21 as a search condition, and acquires the estimated remaining rate 31. For example, in the configuration of the maintenance support system 1 shown in FIG. 1, the terminal 2 receives the selection of the prediction target 21 and transmits it to the server 3. The server 3 searches the DB 5 using the received prediction target 21 as a search condition, and transmits the estimated remaining rate 31 to the terminal 2.

  The estimated remaining rate 31 reflects, for example, the result of a periodic security survey that is performed by periodically visiting a general household where the equipment is installed. As a result, the maintenance support system 1 can provide maintenance support information based on accurate grasp of the actual situation.

  The actual installed number 32 is the number of devices sold or installed in the past for each elapsed period. For example, in the configuration of the maintenance support system 1 shown in FIG. 1, the terminal 2 receives the selection of the prediction target 21 and transmits it to the server 3. The server 3 searches the DB 5 using the prediction target 21 to be received as a search condition, totals the number of sold devices for each elapsed period, and calculates the actual installed number 32. Then, the server 3 transmits the actual installed number 32 to the terminal 2.

  The actual number of repairs 33 is the number of cases at the time of occurrence of actual device repair. For example, in the configuration of the maintenance support system 1 shown in FIG. 1, the terminal 2 receives the selection of the prediction target 21 and transmits it to the server 3. The server 3 searches the DB 5 using the received prediction target 21 as a search condition, totals the number of repairs of the device for each occurrence point (for example, monthly), and calculates the actual repair number 33. Then, the server 3 transmits the actual repair number 33 to the terminal 2.

  The estimated remaining number 41, the estimated hazard function 42, and the estimated cumulative repair rate 43 are data calculated by the estimated remaining number calculating process 62, the estimated hazard function calculating process 63, and the estimated cumulative repair rate calculating process 64, respectively. In particular, these data are data directly derived from the estimated value and the actual value.

  The estimated remaining number 41 is an estimated value of the number of units for each elapsed period in which devices are actually remaining. The estimated remaining number calculating process 62 calculates the estimated remaining number 41 for each elapsed period based on the estimated remaining rate 31 and the actual installed number 32. Details of the estimated remaining number calculating process 62 will be described later with reference to FIG.

  The estimated hazard function 42 is an estimated value of the hazard function of the device. “Hazard function” is a statistical term. In the embodiment of the present invention, the “hazard function” indicates the probability (failure rate) of failure at time t <x <t + Δt under the assumption that the system is operating normally until time t.

  The estimated hazard function calculation process 63 calculates the estimated hazard function 42 based on the estimated remaining number 41 and the actual repair number 33. Details of the estimated hazard function calculation process 63 will be described later with reference to FIG.

  The estimated cumulative repair rate 43 is an estimated value of the cumulative repair rate of the equipment. The estimated cumulative repair rate calculation process 64 calculates an estimated cumulative repair rate 43 based on the estimated hazard function 42. Details of the estimated remaining number calculating process 62 will be described later with reference to FIG.

  The predicted cumulative repair rate 51, the predicted remaining number 52, the predicted hazard function 53, the predicted repair rate 54, and the predicted repair number 55 (hereinafter collectively referred to as “predicted result 50”) are predicted processing 65a, 65b, 65c, respectively. Data calculated by 65d and 65e (hereinafter, collectively referred to as “prediction process 65”). In particular, these data are calculated by performing a fitting using the prediction model 22 and predicting data in a range where there is no actual value or a range where there is an actual value but is not used.

  The predicted cumulative repair rate 51 is a predicted value of the future cumulative repair rate of the device. When the prediction range is set in the past, the predicted cumulative repair rate 51 is a predicted value of the past cumulative repair rate of the device. Although details of the case where the prediction range is set in the past will be described later, it is possible to compare the calculated prediction value with the actual value by setting the prediction range in the past. The meaning when the prediction range is set in the past is the same for the predicted remaining number 52, the predicted hazard function 53, the predicted repair rate 54, and the predicted repair number 55.

  The prediction process 65 a calculates the predicted cumulative repair rate 51 by applying the prediction model 22 to the estimated cumulative repair rate 43 using data within the range of the learning period 23. Details of the prediction process 65a will be described later with reference to FIG.

  The predicted remaining number 52 is a predicted value of the future remaining number of devices. The prediction process 65 b calculates the predicted remaining number 52 based on the estimated remaining number 41 and the predicted cumulative repair rate 51. Details of the prediction process 65b will be described later with reference to FIG.

  The predicted hazard function 53 is a predicted value of the future hazard function of the device. The prediction process 65 c calculates a predicted hazard function 53 based on the predicted cumulative repair rate 51. Details of the prediction process 65c will be described later with reference to FIG.

  The predicted repair rate 54 is a predicted value of the future repair rate of the device. The prediction process 65 d calculates a predicted repair rate 54 based on the predicted cumulative repair rate 51. Details of the prediction process 65d will be described later with reference to FIG.

  The predicted number of repairs 55 is a predicted value of the number of future repairs of the device. The prediction process 65 e calculates a predicted repair number 55 based on the predicted remaining number 52, the predicted hazard function 53, and the predicted repair rate 54. Details of the prediction process 65e will be described later with reference to FIG.

  The graph data 56 is data for displaying the prediction result 50 that is the result of the prediction process 65 in a graph. The result output process 66 outputs graph data 56 based on the prediction result 50. The output destination is, for example, the storage unit 12, the media input / output unit 13, the communication control unit 14, the display unit 16, the peripheral device I / F unit 17, and the like.

  When the terminal 2 outputs the graph data 56 to the display unit 16 or outputs the graph data 56 to a printer or the like via the peripheral device I / F unit 17 or the like, the user can check the graph data 56. Then, the user determines whether or not to cause the maintenance support system 1 to execute the prediction process again while checking the graph data 56. When the prediction process is executed again, the user selects the prediction target 21, the prediction model 22, the learning period 23, and the like via the input unit 15 of the terminal 2.

  4 to 7 show details of data registered in the DB 5. For convenience of explanation, the table is divided into four tables, but may be combined into one table or more tables. 4 to 7 merely show simplified data necessary for describing the embodiment of the present invention, and more detailed data may be used.

  FIG. 4 is an example of the device installation information 70. The device installation information 70 is information indicating the customer who installed the device in the past, the installation time, and the like. As illustrated in FIG. 4, the device installation information 70 uniquely identifies, for example, No 71 that uniquely identifies data, a device code 72 that uniquely identifies the device model number, a model 73 that indicates the device type, and a customer. Customer code 74, installation date 75 indicating the date of installation, etc. are included. Instead of the model 73, a model code that uniquely identifies the type of device may be used.

  FIG. 5 is an example of the estimated remaining rate information 80. The estimated remaining rate information 80 is information indicating the estimated remaining rate 31. As shown in FIG. 5, the estimated remaining rate information 80 includes, for example, No 81 that uniquely identifies data, a model 82 that indicates the type of device, an elapsed period 83 that indicates an elapsed time since the installation of the device, and an estimated remaining rate 31 Etc. Instead of the model 82, a model code for uniquely identifying the type of device may be used.

  FIG. 6 is an example of the repair information 90. The repair information 90 is information indicating the contents of repair of the device performed in the past. As shown in FIG. 6, the repair information 90 indicates, for example, No 91 that uniquely identifies data, a device code 92 that uniquely identifies the device model number, a model 93 that indicates the device type, and the date on which the device was repaired. Repair date 94, failure phenomenon 95 indicating device failure phenomenon, repair location 96 indicating device repair location, failure cause 97 indicating device failure cause, repair method 98 indicating device repair method, repair by parts replacement A part code 99 or the like indicating the replaced part at the time of the change is included. Instead of the model 93, a model code for uniquely identifying the type of device may be used. Similarly, the failure phenomenon 95, the repair location 96, the failure cause 97, and the repair method 98 may be coded.

  FIG. 7 is an example of the maintenance execution information 100. The maintenance execution information 100 is information indicating the content of maintenance related to devices that have been performed in the past. As shown in FIG. 7, the maintenance execution information 100 includes, for example, No 101 for uniquely identifying data, a device code 102 for uniquely identifying the model number of the device, a maintenance execution month 103 indicating when the maintenance was performed, The maintenance execution contents 104 indicating the execution contents are included. The maintenance execution content 104 may be coded.

  FIG. 8 is a flowchart showing a processing flow of the maintenance support system 1. The maintenance support system 1 performs various data processing processes. The process shown in FIG. 8 is a part of main processes in the maintenance support system 1.

  As illustrated in FIG. 8, the control unit 11 of the terminal 2 accepts selection of a prediction target 21, a prediction model 22, and a learning period 23 (S1). That is, when the user selects the prediction target 21, the prediction model 22, and the learning period 23 via the input unit 15, the control unit 11 of the terminal 2 selects the selected prediction target 21, the prediction model 22, and the learning period. 23 is input as data used for processing to be described later.

  The control unit 11 of the terminal 2 acquires the estimated remaining rate 31, the actual installed number 32, and the actual repair number 33 based on the prediction target 21 (S2).

  First, an example of the process for obtaining the estimated remaining rate 31 will be described. The control unit 11 of the terminal 2 transmits the device code selected as the prediction target 21 to the server 3. The control unit 11 of the server 3 identifies the model from the device code received from the terminal 2. The server 3 specifies the model by searching for basic device information (not shown) (information stored in association with at least the device code and the model). Next, the control unit 11 of the server 3 searches the estimated remaining rate information 80 using the specified model as a search condition. Then, the control unit 11 of the server 3 transmits to the terminal 2 list data of the elapsed period 83 and the estimated remaining rate 84 of the estimated remaining rate information 80 obtained as a search result. For example, when the estimated remaining rate information 80 is registered in units of months, the list data of the elapsed period 83 and the estimated remaining rate 84 is also in units of months. The data transmitted to the terminal 2 is the estimated remaining rate 31 for each elapsed period (monthly unit).

  Next, an example of the process for acquiring the actual installed number 32 will be described. The control unit 11 of the server 3 searches the device installation information 70 using the device code received from the terminal 2 as a search condition. And the control part 11 of the server 3 totals the data obtained as a search result, for example per month using the installation date 75. FIG. Then, the control unit 11 of the server 3 transmits the monthly data obtained as the totaling result to the terminal 2. The data transmitted to the terminal 2 is the actual installed number 32 for each elapsed period (monthly).

  Next, an example of an acquisition process of the actual repair number 33 will be described. The control unit 11 of the server 3 searches the repair information 90 using the device code received from the terminal 2 as a search condition. And the control part 11 of the server 3 totals the data obtained as a search result, for example per month using the repair date 94. Then, the control unit 11 of the server 3 transmits the monthly data obtained as the totaling result to the terminal 2. The data transmitted to the terminal 2 is the actual number of repairs 33 for each occurrence point (monthly).

  Next, the control unit 11 of the terminal 2 calculates the estimated remaining number 41 based on the estimated remaining rate 31 and the actual installed number 32 (S3).

  An example of the calculation process of the estimated remaining number 41 will be described. For example, let us consider a case where the estimated remaining number 41 as of January 2011 is calculated for a device sold since January 2001. When the actual installed number 32 in January 2001 is 10,000, and the estimated remaining rate 31 in the elapsed period (monthly) is 10 years, the control unit 11 of the terminal 2 is as of January 2011. It is estimated that 10,000 units × 50% = 5,000 units among the 10,000 units installed in January 2001 remain. Similarly, if the actual installed number 32 in February 2001 is 20,000, the estimated remaining rate 31 is 9% and 11 months and the elapsed period (monthly) is 55%, as of January 2011, 2001 It is estimated that 20,000 units × 55% = 11,000 units remain in 20,000 units installed in February. Thereafter, the same processing is repeated until December 2010, and the estimated remaining number 41 as of January 2011 is calculated by obtaining the sum of all the numbers. In addition to the elapsed period, the control unit 11 of the terminal 2 further estimates the number of remaining units for each installation year, obtains the total number of those units, and obtains the estimated remaining number 41 as of January 2011. calculate.

  Next, the control unit 11 of the terminal 2 calculates the estimated hazard function 42 based on the estimated remaining number 41 and the actual repair number 33 (S4).

  An example of the calculation process of the estimated hazard function 42 will be described. In the embodiment of the present invention, a case in which a failure occurs without being repaired is removed, and the failure rate is not equal to the repair rate. In the embodiment of the present invention, one feature is that a case that is removed without being repaired even if a failure occurs is added in the form of an estimated remaining rate. In general, the repair rate is determined by repair rate = number of repairs / number of actual installed units (= initial number) for each elapsed period. However, the number of repairs here means the number of cases where there is no equipment removal on the way. Therefore, when there is a removal in the middle (or when the removal in the middle is considered), the repair rate cannot be calculated directly. Therefore, in the embodiment of the present invention, first, a hazard function is calculated, then the cumulative repair rate is estimated from the hazard function, and finally the repair rate is obtained from the difference of the cumulative repair rate. Therefore, for example, the control unit 11 of the terminal 2 first calculates the estimated hazard function 42 = the actual number of repairs 33 / (estimated remaining number 41 × (1-cumulative repair rate up to the previous period)) for each elapsed period (monthly). Is calculated. The denominator of the hazard function here is the estimated number of devices that remain in a certain period and have not failed by the previous period.

  Next, the control unit 11 of the terminal 2 calculates the estimated cumulative repair rate 43 based on the estimated hazard function 42 (S5).

  An example of the calculation process of the estimated cumulative repair rate 43 will be described. Since the estimated hazard function 42 is the probability that a device that was healthy until the previous period for each elapsed period (monthly) will break down, the control unit 11 of the terminal 2 determines the value of each estimated period of time (monthly) for the estimated hazard function 42. The estimated cumulative repair rate 43 is calculated sequentially. For example, the estimated cumulative repair rate at the 120th month is 1- (estimated hazard function at the 1-1 month) × (estimated hazard function at the 1-2 month) ×. Hazard function).

  Next, the control unit 11 of the terminal 2 calculates a predicted cumulative repair rate 51 based on the estimated cumulative repair rate 43, the prediction model 22, and the learning period 23 (S6).

  An example of the calculation process of the predicted cumulative repair rate 51 will be described. The control unit 11 of the terminal 2 calculates the predicted cumulative repair rate 51 by applying the prediction model 22 to the estimated cumulative repair rate 43 using data within the range of the learning period 23. The data within the range of the learning period 23 is data in which the elapsed period (monthly) is included within the range of the learning period 23 in the data group of the estimated cumulative repair rate 43 calculated in S5. Examples of the prediction model 22 include the Weibull model and the linear model described above. The control unit 11 of the terminal 2 calculates the predicted cumulative repair rate 51 of the prediction period by applying the data of the estimated cumulative repair rate 43 included in the range of the learning period 23 to the prediction model 22 by a known statistical process. The prediction period is a range in which there is no actual value, that is, the future unless otherwise specified by the user.

  Next, the control unit 11 of the terminal 2 calculates the predicted remaining number 52, the predicted hazard function 53, and the predicted repair rate 54 based on the predicted cumulative repair rate 51 and the estimated remaining number 41 (S7).

  First, an example of the calculation process of the predicted remaining number 52 will be described. For example, the control unit 11 of the terminal 2 calculates the predicted remaining number 52 using the estimated remaining rate 31 used for the calculation process of the estimated remaining number 41.

  Next, an example of calculation processing of the predicted hazard function 53 and the predicted repair rate 54 will be described. The process for obtaining the predicted hazard function 53 from the predicted cumulative repair rate 51 is the reverse of the process for obtaining the estimated cumulative repair rate 43 from the estimated hazard function 42. That is, the control unit 11 of the terminal 2 calculates the predicted hazard function 53 by using the predicted cumulative repair rate 51 between adjacent months. For example, the predicted hazard function for the 130th month can be calculated by (predicted cumulative repair rate for the 1-130th month) / (predicted cumulative repair rate for the 1-129th month). Further, the difference value of the predicted cumulative repair rate 51 between adjacent months is the predicted repair rate 54 for each elapsed period (monthly unit).

  Next, the control unit 11 of the terminal 2 calculates the predicted repair number 55 based on the predicted remaining number 52 and the predicted hazard function 53 or the predicted repair rate 54 (S8).

  An example of a process for calculating the predicted number of repairs 55 will be described. The process for obtaining the predicted number of repairs 55 is based on the predicted remaining number 52 for each elapsed month (month) at the time of occurrence (monthly) to be predicted (1—predicted cumulative repair rate 51 up to the previous period or estimated cumulative repair rate 43). Multiply the predicted number of occurrences you want to predict by multiplying by the predicted hazard function 53 or estimated hazard function 42 for each elapsed period (monthly), and take the total for all elapsed years (monthly) Thus, the predicted number of repairs 55 at the time of occurrence (monthly) to be predicted is calculated. In principle, the estimated cumulative repair rate 43 and the estimated hazard function 42 are used for the elapsed period with actual results, and the predicted cumulative repair rate 51 and the predicted hazard function 53 are used for the other periods. Alternatively, the predicted cumulative repair rate 51 and the predicted hazard function 53 may be used for smoothing. For example, if the predicted number of repairs as of January 2011 is calculated, the number of predicted repairs = (predicted hazard function or estimated hazard function for an elapsed period of one month × predicted remaining number of months for an elapsed period of one month as of January 2011 × ( 1−Estimated cumulative repair rate or estimated cumulative repair rate for 0 month elapsed time + (Estimated hazard function or estimated hazard function for 2 months elapsed time × Predicted remaining number of 2 months elapsed as of January 2011 × (1− (Estimated cumulative repair rate or estimated cumulative repair rate for an elapsed period of one month) +... (Take the total up to the converged elapsed period).

  Next, the control unit 11 of the terminal 2 displays various graphs (S9). For example, the control unit 11 of the terminal 2 sets the prediction result 50 (the predicted cumulative repair rate 51, the predicted remaining number 52, the predicted hazard function 53, the predicted repair rate 54, the predicted repair number 55) and the corresponding past results to be the same. By controlling to display in the graph, the user can visually confirm the validity of the prediction. After confirming, the user can reselect the prediction target 21, the prediction model 22, and the learning period 23 as necessary, and adjust the prediction so that the prediction is good.

  With the processing shown in FIG. 8, the maintenance support system 1 can provide accurate information to the user as maintenance support information for making a maintenance plan for the device. In particular, if the estimated remaining rate stored in the DB 5 reflects the result of a periodic security investigation that periodically goes around a general home where the equipment is installed, the maintenance support system 1 Maintenance support information based on accurate grasp of the actual situation can be provided to the user.

  Further, the maintenance support system 1 accepts the selection of the learning period 23, applies the prediction model 22 to the estimated cumulative repair rate 43 using the data within the range of the learning period 23, and calculates the prediction result 50. Since the prediction process is performed, the user can dynamically change the learning period and obtain maintenance support information. Therefore, the user can select an appropriate learning period according to past cases (for example, cases such as breakdowns and repairs) and the like, and can obtain accurate maintenance support information.

  In addition to the predicted cumulative repair rate 51, the maintenance support system 1 calculates the predicted remaining number 52, the predicted hazard function 53, the predicted repair rate 54, and the predicted repair number 55. Various information can be obtained.

  Below, each modification in embodiment of this invention is demonstrated.

<Modification 1>
In the first modification, the maintenance support system 1 displays the maintenance execution information 100 and displays the selectable range of the learning period 23. FIG. 9 is a display example of the selectable range of the maintenance execution information 100 and the learning period 23.

  In FIG. 9, the horizontal axis represents the elapsed period (elapsed months). This indicates that the starting point is the point in time when the actual device is installed (hereinafter referred to as “the actual point in time”), and now 120 months have passed since the “in-place point”. The period before the current time is a period of “with actual value”. On the other hand, the period after the current time point is the “prediction period”.

  In FIG. 9, “preliminary maintenance start” is displayed as the maintenance execution information 100 when 60 months have elapsed from “the actual installation time”. This indicates that pre-maintenance has been started for the prediction target device when 60 months have passed since the “actual installation time”.

  As an example of narrowing down the learning period, (1) for a certain device, when knowledge is found that “abrupt deterioration starts from about the 60th month”, prediction is performed using only data from the 60th month onward. And (2) when all the preventive measures are taken in the inspection at the 60th month from the installation of a certain device, the prediction is made using the data after the 60th month. .

  Therefore, the maintenance support system 1 may have a function of limiting the selectable range of the learning period 23 based on the maintenance execution information 100, for example. Thereby, the maintenance support system 1 can make it impossible to select a period that should not be included in the learning period from past cases. For example, even if parts replacement for all sold devices is performed and the tendency of occurrence of failure or repair has changed significantly, the maintenance support system cannot select a period before the part replacement time. By doing so, accurate maintenance support information can be calculated.

  Further, the maintenance support system 1 may simply display the maintenance execution information 100 on the display unit 16 of the terminal 2. Thus, the user can select an appropriate learning period while confirming the maintenance execution information 100.

<Modification 2>
In the second modification, the maintenance support system 1 automatically determines the learning period 23 based on the maintenance execution information 100. FIG. 10 is a flowchart showing the process of the second modification. Hereinafter, Modification 2 will be described with reference to FIG. Detailed description of processing similar to that in the flowchart of FIG. 8 is omitted.

  The control unit 11 of the terminal 2 receives selection of the prediction target 21 and the prediction model 22 (S11). The process of S11 is the same as the process of S1 of FIG.

  Next, the control part 11 of the terminal 2 determines the optimal learning period 23 based on the maintenance execution information 100 (S12).

  An example of the determination process for the learning period 23 will be described. As a premise, the storage unit 12 of the terminal 2 stores a maintenance implementation content 104 (hereinafter referred to as “restricted maintenance implementation content”) that restricts the selectable range of the learning period 23 and is designated in advance by the user. To do. For example, when the user wants to limit the selectable range of the learning period 23 for “preliminary maintenance start” and “part replacement”, the “preliminary maintenance start” and “part replacement” are set as the limited maintenance execution contents, and the storage unit of the terminal 2 12 is stored.

  Then, the control unit 11 of the terminal 2 transmits the device code of the prediction target 21 to the server 3. The control unit 11 of the server 3 searches the maintenance execution information 100 using the device code received from the terminal 2 as a search condition. The control unit 11 of the server 3 transmits the maintenance execution month 103 and the maintenance execution content 104 obtained as a search result to the terminal 2. The control unit 11 of the terminal 2 compares the maintenance execution content 104 received from the server 3 with the limited maintenance execution content stored in the storage unit 12. The period from the point in time to the end point with the actual value is determined as the learning period 23. The reason for determining the learning period 23 in this way is that, with the exception of data with different tendency of occurrence of failure and repair, it is considered that the more data used for the prediction process 65, the higher the prediction accuracy.

  Next, the control unit 11 of the terminal 2 acquires the actual installed number 32 and the actual repair number 33 (S13), calculates the estimated remaining number 41, the estimated hazard function 42, and the estimated cumulative repair rate 43 (S14), The prediction process 65 is executed using the learning period 23 automatically determined in S12 (S15), and the prediction result 50 is output (S16). The processes of S13 to S16 are the same as the processes of S2 to S9 in FIG.

  With the processing shown in FIG. 10, the maintenance support system 1 can save the user's trouble and can provide accurate maintenance support information to a user who is not familiar with the system. For example, when parts replacement is performed for all sold devices and the tendency of occurrence of failure or repair has changed significantly, the maintenance support system should determine the period after the part replacement time as the learning period Thus, accurate maintenance support information can be calculated.

<Modification 3>
In Modification 3, the maintenance support system 1 performs a plurality of prediction processes and compares and outputs the prediction results. FIG. 11 is a flowchart showing the process of the third modification. FIG. 12 is an output example of the graph data 56. Hereinafter, Modification 3 will be described with reference to FIG. Detailed description of processing similar to that in the flowchart of FIG. 8 is omitted.

  The control unit 11 of the terminal 2 receives selection of the prediction target 21 and the prediction model 22 (S21). The process of S21 is the same as the process of S1 of FIG.

  Next, the control unit 11 of the terminal 2 accepts selection of a plurality of learning periods 23 (S22). Note that the control unit 11 of the terminal 2 may accept selection of a plurality of prediction models 22.

  Next, the control unit 11 of the terminal 2 acquires the actual installed number 32 and the actual repair number 33 (S23), and calculates the estimated remaining number 41, the estimated hazard function 42, and the estimated cumulative repair rate 43 (S24). The processes of S23 and S24 are the same as the processes of S2 to S5 in FIG.

  Next, the control part 11 of the terminal 2 performs the prediction process 65 using one of the learning periods 23 selected in S12 (S25). The process of S25 is the same as the process of S6 to S8 in FIG.

  Next, the control part 11 of the terminal 2 determines whether the prediction process 65 was complete | finished about all the learning periods 23 (S26). If not completed (No in S26), the control unit 11 of the terminal 2 repeats the process of S25. When it has been completed (Yes in S26), the control unit 11 of the terminal 2 compares and outputs a plurality of prediction results 50 (S27).

  FIG. 12A is a graph of the predicted cumulative repair rate 51 of the device A as of 2005. The learning period has an elapsed period (monthly) of 24 to 60 months, and an occurrence point (monthly) of 2000 to 2005. The prediction period is an elapsed period (monthly) after 60 months.

  FIG. 12B is a graph of the predicted cumulative repair rate 51 of the device A as of 2010. The learning period has an elapsed period (monthly) of 60 to 120 months, and an occurrence time (monthly) of 2006 to 2010. The prediction period is an elapsed period (monthly) after 120 months.

  FIG. 12C is a graph in which the predicted cumulative repair rate 51 of the device A as of 2005 and the predicted cumulative repair rate 51 as of 2010 are compared and output in the same graph.

  In the example shown in FIG. 12, prior maintenance has started for the device A in 2005, and the tendency of occurrence of failures and repairs has largely changed between before and after 2005. However, as shown in FIG. 12A and FIG. 12B, it is difficult to read such a tendency with the result output for each time. On the other hand, by comparing and outputting as shown in FIG. 12C, the difference between the two can be visually determined. Therefore, even if the trend change in 2005 is unknown, the user performs maintenance work that causes a trend change in the past by confirming the comparison output as shown in FIG. I can find out.

  The process shown in FIG. 11 allows the user to compare the prediction results from a plurality of learning periods. Therefore, the user can find out that maintenance work that causes a trend change in the past has been performed.

<Modification 4>
In the modified example 4, the maintenance support system 1 calculates the degree of deviation of the prediction process 65 and determines whether re-prediction is necessary. FIG. 13 is a flowchart showing the process of the fourth modification. Hereinafter, Modification 4 will be described with reference to FIG. Detailed description of processing similar to that in the flowchart of FIG. 8 is omitted.

  The control unit 11 of the terminal 2 receives selection of a past prediction period (S31). The past prediction period means a period in which there is a record value and it is not necessary to predict it. In the modified example 4, in order to verify the validity of the prediction model 22 and the learning period 23 used in the prediction process 65, the prediction process 65 is executed by selecting a past prediction period and compared with the actual value.

  Next, the control part 11 of the terminal 2 receives selection of the prediction object 21, the prediction model 22, and the learning period 23 (S32). The process of S32 is the same as the process of S1 of FIG.

  Next, the control unit 11 of the terminal 2 acquires the actual installed number 32 and the actual repair number 33 (S33), calculates the estimated remaining number 41, the estimated hazard function 42, and the estimated cumulative repair rate 43 (S34), The prediction process 65 for the past prediction period is executed (S35). The processing of S33 to S35 is the same as the processing of S2 to S8 in FIG.

  Next, the control unit 11 of the terminal 2 acquires the actual value of the same period as the past prediction period from the DB 5 of the server 3 (S36), and calculates the degree of deviation (S37).

  An example of the deviation degree calculation process will be described. The degree of deviation is the difference between the predicted value and the actual value, and in terms of statistics, it is an “error” between the predicted value and the actual value. For example, when calculating the degree of deviation of the predicted cumulative repair rate 51, the control unit 11 of the terminal 2 calculates the mean square error between the calculated predicted cumulative repair rate 51 and the actual cumulative repair rate in the same period as the past predicted period. The sum of is calculated.

  Next, the control unit 11 of the terminal 2 determines whether or not the degree of detachment is greater than or equal to a predetermined threshold (S38). The predetermined threshold value is stored in the storage unit 12 in advance.

  When the degree of detachment is equal to or greater than the predetermined threshold (Yes in S38), the control unit 11 of the terminal 2 repeats the process from S32. When the degree of deviation is less than the predetermined threshold (No in S38), the control unit 11 of the terminal 2 calculates a confidence interval of the prediction result 50 by a known statistical process (S39), and together with the confidence interval of the prediction result 50 The prediction result 50 is output (S40).

  In the process illustrated in FIG. 13, when the past prediction period is accepted, the maintenance support system 1 performs the prediction process 65 for the past prediction period, and the past prediction value calculated by the prediction process 65 and the past Are compared with the actual value in the same period, and the degree of deviation from the actual value is calculated. As a result, the user can confirm the degree of deviation from the actual value. Therefore, the user can verify the validity of the prediction model 22 and the learning period 23 used for the prediction process 65.

  In the process shown in FIG. 13, the maintenance support system 1 determines whether or not the calculated degree of detachment is equal to or greater than a predetermined threshold. If the degree of detachment is determined to be equal to or greater than the predetermined threshold, the learning period If the reselection of 23 is accepted and it is determined that the degree of detachment is less than a predetermined threshold, the confidence interval of the prediction result 50 is calculated. This eliminates the need for the user himself to change the learning period 23 or the like and further determine whether or not the maintenance support system 1 should execute the prediction process 65. Furthermore, the user can confirm the confidence interval of the prediction result 50. Therefore, the maintenance support system 1 can give the user a sense of satisfaction with the maintenance support information.

<Modification 5>
In Modification 5, the maintenance support system 1 determines the optimal prediction model 22 and learning period 23 by brute force. FIG. 14 is a flowchart showing the process of the fifth modification. Hereinafter, Modification 5 will be described with reference to FIG. Detailed description of processing similar to that in the flowchart of FIG. 8 is omitted.

  The control unit 11 of the terminal 2 accepts selection of the prediction target 21 (S51). Next, the control unit 11 of the terminal 2 acquires all possible prediction models 22 and learning periods 23 (S52).

  Next, the control unit 11 of the terminal 2 acquires the estimated remaining rate 31, the actual installed number 32, and the actual repair number 33 (S53), and calculates the estimated remaining number 41, the estimated hazard function 42, and the estimated cumulative repair rate 43. Then, the prediction process 65 for the prediction period is executed (S55). The prediction period is a past prediction period in which the actual value exists as in the fourth modification. The processing of S53 to S55 is the same as the processing of S2 to S8 in FIG.

  Next, the control unit 11 of the terminal 2 acquires the actual value of the same period as the prediction period from the DB 5 of the server 3 (S56), and calculates the sum of the mean square error with the actual value (S57).

  Next, the control unit 11 of the terminal 2 determines whether or not the processing has been completed for all the prediction models 22 and the learning periods 23 (S58). When the process has not ended (No in S58), the control unit 11 of the terminal 2 repeats the process from S52. When the process has been completed (Yes in S58), the control unit 11 of the terminal 2 determines the prediction model 22 and the learning period 23 with the smallest sum of mean square errors as the optimum values (S59), and performs the optimum prediction. The model 22 and the learning period 23 are output (S60).

  In the process illustrated in FIG. 14, the maintenance support system 1 determines the learning period 23 one by one from all possible periods, repeats the prediction process 65, and determines the optimum learning period 23. Thereby, the maintenance support system 1 can perform the prediction process 65 for all possible periods, and can automatically determine the optimal learning period 23 from the prediction result 50.

<Modification 6>
In the modified example 6, the maintenance support system 1 uses repair information 90 including a failure phenomenon 95, a failure cause 97, a repair location 96, and a repair method 98 as a data range narrowing condition used for the prediction process 65. That is, the maintenance support system 1 accepts selection of one or a plurality of values of the failure phenomenon 95, the failure cause 97, the repair location 96, and the repair method 98, and selects the data narrowed down by the narrowing-down conditions of the prediction target 21. The prediction process 65 is performed using the same.

  The narrowing-down process in the modified example 6 is included in the process for accepting selection of the prediction target 21. Therefore, the narrowing-down process in the modified example 6 can be applied to all the flowcharts of FIGS. 8, 10, 11, 13, and 14 described above.

  By the narrowing-down process in the modified example 6, the user selects any one or more values of the failure phenomenon 95, the failure cause 97, the repair location 96, and the repair method 98 to narrow down the data range to obtain prediction support information. be able to. As a result, the user can obtain various prediction support information according to the purpose of the maintenance work.

  For example, when it is desired to obtain prediction support information for each failure phenomenon 95 for a specific device, the modified example 6 may be applied to the modified example 3. Specifically, the maintenance support system 1 selects a plurality of failure phenomena 95 (for example, selects “all” of failure phenomena 95 instead of accepting selection of a plurality of learning periods 23 in S22 of FIG. 11. ), And after performing the narrowing-down process in the modified example 6 for each failure phenomenon 95, the prediction process 65 may be executed.

  In addition, for example, when it is desired to execute the prediction process 65 by assigning priorities in the narrowing-down conditions of the prediction target 21 to a specific device, and to obtain prediction support information in which the degree of deviation is less than a predetermined threshold value, Modification 6 may be applied to Modification 4. Specifically, in S32 of FIG. 13, selection of the priority order of the narrowing-down conditions (failure phenomenon 95, failure cause 97, repair location 96, and repair method 98) of the prediction target 21 is accepted, and the modified example 6 according to the priority order. After performing the narrowing-down process in, the prediction process 65 may be executed.

  The preferred embodiments of the maintenance support system and the like according to the present invention have been described above with reference to the accompanying drawings, but the present invention is not limited to such examples. It will be apparent to those skilled in the art that various changes or modifications can be conceived within the scope of the technical idea disclosed in the present application, and these naturally belong to the technical scope of the present invention. Understood.

1 ... Maintenance support system 2 ......... Terminal 3 ... Server 4 ... Maintenance support program 5 ... Database (DB)
6 ……… Network 21 ……… Predicted object 22 ……… Predicted model 23 ……… Learning period 31 ……… Estimated remaining rate 32 ……… Actual installation number 33 ……… Actual repair number 41 ……… Estimated remaining Number of units 42 ………… Estimated hazard function 43 ……… Estimated cumulative repair rate 50 ……… Predicted result 51 ……… Predicted cumulative repair rate 52 ……… Predicted remaining number of units 53 ……… Predicted hazard function 54 ……… Predicted repair Rate 55 ......... Estimated number of repairs 56 ... ... Graph data 61a, 61b, 61c ... ... Acquisition process 62 ... ... Estimated remaining number calculation process 63 ... ... Estimated hazard function calculation process 64 ... ... Estimated cumulative repair rate Calculation processing 65, 65a, 65b, 65c, 65d, 65e ......... Prediction processing 66 ......... Result output processing

Claims (16)

A maintenance support system configured by a computer and supporting maintenance work of equipment,
Storage means for storing an estimated remaining rate that is an estimated value of a rate at which the device is actually operating for each elapsed period from the installation time of the device;
An acquisition means for acquiring the actual number of installed devices for each elapsed period, and acquiring the actual number of repairs of the devices for each occurrence time;
Based on the estimated remaining rate and the actual installed number of units, an estimated remaining number calculating means for calculating an estimated remaining number that is an estimated value of the number of devices actually remaining for each elapsed period;
An estimated hazard function calculating means for calculating an estimated hazard function that is an estimated value of the hazard function of the device based on the actual repair number and the estimated remaining number of units;
Based on the estimated hazard function, an estimated cumulative repair rate calculating means for calculating an estimated cumulative repair rate that is an estimated value of the cumulative repair rate of the equipment;
A maintenance support system comprising: Accepting means for accepting selection of a learning period;
A prediction process that uses a data within the range of the learning period to apply a predetermined prediction model to the estimated cumulative repair rate and calculates a predicted cumulative repair rate that is a predicted value of a future cumulative repair rate of the device Prediction means for performing
The maintenance support system according to claim 1, further comprising: In the prediction process, a predicted remaining number that is a predicted value of the future remaining number of the device, a predicted hazard function that is a predicted value of the future hazard function of the device, and a predicted value of a future repair rate of the device The maintenance support system according to claim 2, wherein a predicted repair rate and a predicted repair number that is a predicted value of a future repair number of the device are calculated. The storage means further stores maintenance execution information regarding the device for each occurrence time point,
4. The maintenance support system according to claim 2, further comprising display means for displaying the maintenance execution information. The maintenance support system according to claim 4, wherein the reception unit further has a function of limiting a selectable range of the learning period based on the maintenance execution information. The maintenance support system according to claim 4, wherein the reception unit further has a function of automatically determining the optimal learning period based on the maintenance execution information. The accepting means accepts selection of a plurality of the learning periods;
The maintenance support system according to claim 2, wherein the prediction unit performs a plurality of the prediction processes for each of the plurality of learning periods and compares and outputs a prediction result. The accepting means accepts selection of a past prediction period;
The prediction means performs the prediction process for the past prediction period,
A deviation degree calculating means for comparing a prediction value for the past prediction period calculated by the prediction process with an actual value in the same period as the past prediction period, and calculating a deviation degree from the actual value; The maintenance support system according to claim 2, further comprising: Determination means for determining whether or not the detachment degree calculated by the detachment degree calculation means is greater than or equal to a predetermined threshold;
When the determination means determines that the degree of detachment is equal to or greater than a predetermined threshold, the reception means accepts re-selection of the learning period,
The maintenance support system according to claim 8, wherein the prediction unit calculates a confidence interval of a prediction result when the determination unit determines that the degree of detachment is less than a predetermined threshold. The prediction means further has a function of determining the learning period one by one from all possible periods, repeating the prediction process, and determining the optimum learning period. Item 10. The maintenance support system according to any one of Items 9. The storage means further stores repair information including a failure phenomenon of the device, a cause of failure, a repair location, and a repair method,
The accepting unit further accepts selection of one or more values of the failure phenomenon, the cause of the failure, the repair location, and the repair method as a condition for narrowing the data range used for the prediction process,
11. The maintenance support system according to claim 2, wherein the prediction unit performs the prediction process using data narrowed down according to the narrow-down condition received by the reception unit. The accepting means accepts selection of the prediction model;
12. The maintenance support system according to claim 2, wherein the prediction unit performs the prediction process by applying the prediction model received by the reception unit. A maintenance support method executed by a computer and supporting maintenance work of equipment,
The computer is
A storage step of storing an estimated remaining rate that is an estimated value of a rate at which the device is actually operating for each elapsed period from the installation time of the device;
An acquisition step of acquiring the actual number of installed devices for each elapsed period, and acquiring the actual number of repairs of the devices for each occurrence point;
Based on the estimated remaining rate and the actual installed number of units, for each elapsed period, an estimated remaining number calculating step for calculating an estimated remaining number that is an estimated value of the number of devices actually remaining;
An estimated hazard function calculating step for calculating an estimated hazard function that is an estimated value of the hazard function of the device based on the actual repair number and the estimated remaining number of units;
Based on the estimated hazard function, an estimated cumulative repair rate calculating step for calculating an estimated cumulative repair rate that is an estimated value of the cumulative repair rate of the device;
A maintenance support method comprising: A reception step for accepting selection of a learning period;
A prediction process that uses a data within the range of the learning period to apply a predetermined prediction model to the estimated cumulative repair rate and calculates a predicted cumulative repair rate that is a predicted value of a future cumulative repair rate of the device A prediction step of performing
The maintenance support method according to claim 13, further comprising: A program for causing a computer to execute a maintenance support method for supporting equipment maintenance work,
A storage step of storing an estimated remaining rate that is an estimated value of a rate at which the device is actually operating for each elapsed period from the installation time of the device;
An acquisition step of acquiring the actual number of installed devices for each elapsed period, and acquiring the actual number of repairs of the devices for each occurrence point;
Based on the estimated remaining rate and the actual installed number of units, for each elapsed period, an estimated remaining number calculating step for calculating an estimated remaining number that is an estimated value of the number of devices actually remaining;
An estimated hazard function calculating step for calculating an estimated hazard function that is an estimated value of the hazard function of the device based on the actual repair number and the estimated remaining number of units;
Based on the estimated hazard function, an estimated cumulative repair rate calculating step for calculating an estimated cumulative repair rate that is an estimated value of the cumulative repair rate of the device;
A program for causing a computer to execute a process including: A reception step for accepting selection of a learning period;
A prediction process that uses a data within the range of the learning period to apply a predetermined prediction model to the estimated cumulative repair rate and calculates a predicted cumulative repair rate that is a predicted value of a future cumulative repair rate of the device A prediction step of performing
The program according to claim 15, for causing the computer to execute processing further including:

Images