Fault recovery support system and fault recovery support method

Abstract

To speed up failure recovery by recommending work and components required for recovery when a machine failure occurs.SOLUTION: A failure recovery support system 1 comprises: a work content extraction unit 201 for extracting required work contents from a maintenance document representing a history including documents in which instructed or executed work for recovering failures occurred on apparatuses in the past are recorded; a work-and-components correspondence extraction unit 300 for extracting a work-and-components correspondence in which components of component numbers whose usages are assumed at the time of instructing the work or components of component numbers existing in the history in which component numbers used at the time of executing the work are recorded are assumed to be used or are estimated to have been used correspondingly to which work contents; a recommendation matrix calculation unit 303 for calculating recommendation scores of required work or components in response to input of information identifying a failure occurred on an apparatus; and a recovery unit recommendation unit 6010 for recommending components corresponding to the work calculated by the recommendation matrix calculation unit 303 and work corresponding to the components calculated by the recommendation matrix calculation unit 303 by supplementing the recommendation using the work-and-components correspondence.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to a failure recovery support system and a failure recovery support method for assisting in the recovery of failures in machines composed of a large number of parts (sites) such as wind turbines.

Background Art

[0002] In machines composed of a large number of parts (sites) such as wind turbines, when a failure occurs, an operator performs an operation to recover the failure. And the operator records the results of the maintenance work in a document.

[0003] The invention described in Patent Document 1 provides a technology that can easily create, add, and update an abnormality diagnosis model for initial and new failure cases, and by using that model, can accurately and efficiently realize abnormality diagnosis, countermeasure instructions, etc. The feature of the abnormality diagnosis model described in Patent Document 1 is to create the model by referring to maintenance work data in which the work content related to the device and the used parts are concisely described in a table format, and to recommend the work content and necessary parts according to that model.

[0004] A general form assumed as the history of maintenance work is a document form described in natural language, which does not satisfy the requirements of the maintenance work data assumed in Patent Document 1. In the recommendation using the abnormality diagnosis model of Patent Document 1, the basis of the recommendation is not shown, so it lacks reliability from the operator. The operator cannot grasp the basis of the recommendation only by referring to the abnormality diagnosis model.

Prior Art Documents

Patent Documents

[0005]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0006] The invention described in Patent Document 1 makes it possible to recommend appropriate work procedures and parts to be used for a given failure based on the history of past maintenance work. This allows maintenance planners to comprehensively identify the work and parts necessary for failure recovery. However, a method is needed that can be applied to data in document format, which is generally expected to represent the history of maintenance work. Furthermore, when recommending work and parts, the basis and reliability of the recommendation must be demonstrated by citing the history and examples of past maintenance work, and this must be accepted by the planner.

[0007] Conventional methods can recommend appropriate work procedures and parts to restore a malfunction based on past maintenance work history. However, conventional methods require maintenance work data that concisely describes the work procedures and parts used for the equipment in a tabular format, and have the problem of not being able to handle documents written in natural language. Furthermore, the basis for the recommendations is not shown, which lacks user trust and makes it difficult to decide whether or not to accept the recommendations.

[0008] Therefore, the present invention aims to expedite the recovery process by recommending the necessary work and parts for machine repair in the event of a malfunction. [Means for solving the problem]

[0009] To solve the aforementioned problems, the present invention provides a fault recovery support system that recommends the necessary work procedures and part numbers when a failure occurs in equipment, and records the work that was instructed or performed in the past to recover from failures that occurred in the equipment. nature From the preservation documents, which are historical records including documents, Based on the working word table Work details The system extracts keywords and assigns similar task identifiers to tasks similar to the task in question based on the feature quantities of those keywords. A work content extraction unit extracts work part correspondence relationships by estimating which work content a part with a part number recorded in the history of part numbers that were expected to be used when the work was instructed or when the work was performed corresponds to, and when information identifying a malfunction that has occurred in the equipment is input, the necessary work Similar work identifiers and A recommendation matrix calculation unit that calculates the recommendation score of a component, and the recommendation matrix calculation unit that calculates Similar work identifier The system is characterized by comprising: a component corresponding to the component; and a recovery unit recommendation unit that recommends the work corresponding to the component calculated by the recommendation matrix calculation unit by supplementing it using the work component correspondence relationship. The present invention relates to a fault recovery support system that recommends the necessary work procedures and part numbers when a fault occurs in equipment, comprising: a work procedure extraction unit that extracts the necessary work procedures from maintenance documents, which are a history including documents that record the work instructed or performed to recover from a fault that occurred in the equipment in the past; a work part correspondence relationship extraction unit that extracts a work part correspondence relationship by estimating which work procedures correspond to which parts with part numbers in the history that record the part numbers that were expected to be used when the work was instructed or the part numbers that were used when the work was performed; and information that identifies the fault that occurred in the equipment. The system is characterized by comprising: a recommendation matrix calculation unit that calculates recommendation scores for necessary work or parts when input is received; a recovery unit recommendation unit that recommends parts corresponding to the work calculated by the recommendation matrix calculation unit, and work corresponding to the parts calculated by the recommendation matrix calculation unit, by supplementing them using the work-part correspondence relationship; and a recommendation basis display unit that calculates a reliability score for the maintenance document based on attribute information including the document creator's years of experience, position, same-day repair completion rate, type of maintenance document, and the number of business trips required to recover from the failure, and shows the maintenance document and its reliability score to the user as the basis information for the work or part recommendation.

[0010] The present invention relates to a fault recovery support method that recommends the necessary work procedures and part numbers when a malfunction occurs in equipment, wherein the work procedure extraction unit records one or more instances of work that were instructed or performed in the past to recover from a malfunction in the equipment. nature From the preservation documents, which are a history including documents Based on the working word table Work details The system extracts keywords and assigns similar task identifiers to tasks similar to the task in question based on the feature quantities of those keywords. Steps include: extracting work part correspondence relationships by having the work part correspondence unit extract work part correspondences by estimating which work content a part number from the history of recording the part number that was expected to be used when the work was instructed or the part number that was used when the work was performed corresponds to; and when information identifying a malfunction that occurred in the equipment is input, the necessary work Similar work identifiers and The steps include calculating the recommendation score for the parts, and the recovery unit recommendation unit, which is calculated by the recommendation matrix calculation unit. Similar work identifier The system is characterized by comprising the steps of: complementing and recommending the corresponding parts and the work corresponding to the parts calculated by the recommendation matrix calculation unit using the work part correspondence relationship. The present invention provides a fault recovery support method for recommending necessary work procedures and part numbers when a fault occurs in equipment, comprising: a work procedure extraction unit extracting necessary work procedures from maintenance documents, which are a history containing one or more documents recording the work instructed or performed to recover from a fault that occurred in the equipment in the past; a work part correspondence relationship extraction unit extracting a work part correspondence relationship, which estimates which work procedures a part number in a history recording the part number intended to be used when the work was instructed or the part number used when the work was performed corresponds to; and a recommendation matrix calculation unit calculating the fault that occurred in the equipment The system is characterized by the following steps: when information identifying the work is input, it calculates a recommendation score for the necessary work or parts; the recovery unit recommendation unit complements and recommends the parts corresponding to the work calculated by the recommendation matrix calculation unit, and the work corresponding to the parts calculated by the recommendation matrix calculation unit, using the work-part correspondence relationship; and the recommendation basis display unit calculates a reliability score for the maintenance document based on attribute information including the document creator's years of experience, position, same-day repair completion rate, type of maintenance document, and the number of business trips required to restore the failure, and shows the maintenance document and its reliability score to the user as the basis information for the work or parts recommendation. Other means will be described within the descriptions of embodiments for carrying out the invention. [Effects of the Invention]

[0011] According to the present invention, it is possible to recommend the necessary work and parts for recovery when a machine malfunctions, thereby accelerating the recovery process. [Brief explanation of the drawing]

[0012] [Figure 1] This is a diagram illustrating the fault recovery support system of this embodiment. [Figure 2] This figure shows an example of the data structure of a preservation document table. [Figure 3] It is a diagram showing an example of the data structure of the parts history table. [Figure 4] It is a diagram showing an example of the data structure of the work word table. [Figure 5] It is a diagram showing an example of the data structure of the planner table. [Figure 6] It is a diagram showing an example of the configuration of the work content table. [Figure 7] It is a diagram showing an example of the configuration of the restoration unit table. [Figure 8] It is a diagram showing an example of the configuration of the work parts correspondence table. [Figure 9A] It is a diagram showing an example of the configuration of the failure type parts matrix. [Figure 9B] It is a diagram showing an example of the configuration of the failure type work matrix. [Figure 10] It is a diagram showing an example of the configuration of the parts recommendation matrix. [Figure 11] It is a flowchart of the processing of the work content extraction unit. [Figure 12] It is a flowchart of the processing of the restoration unit extraction unit. [Figure 13] It is a flowchart of the processing of the reliability calculation unit. [Figure 14A] It is a flowchart of the processing of the restoration unit parts estimation unit. [Figure 14B] It is a flowchart of the processing of the restoration unit parts estimation unit. [Figure 14C] It is a flowchart of the processing of the restoration unit parts estimation unit. [Figure 15] It is a flowchart of the processing of the restoration unit aggregation unit. [[ID=]49] [Figure 16A] It is a flowchart of the processing of the recommendation matrix calculation unit. [Figure 16B] It is a flowchart of the processing of the recommendation matrix calculation unit. [Figure 17] It is an example of the restoration unit recommendation screen. [Figure 18] It is a sequence diagram of the processing related to the display of the restoration unit recommendation screen. [Figure 19] It is an example of the recommendation basis screen. [Figure 20] This is a flowchart for processing the basis for recommendations. [Figure 21] This is an example of the preservation document draft screen. [Figure 22] This is a flowchart for processing drafts of maintenance documents. [Modes for carrying out the invention]

[0013] Hereafter, embodiments for carrying out the present invention will be described in detail with reference to the figures. Figure 1 is a configuration diagram showing the fault recovery support system 1 of this embodiment. The fault recovery support system 1 recommends the necessary work procedures and part numbers when a malfunction occurs in equipment such as a wind turbine. The fault recovery support system 1 consists of a business data storage unit 100, a maintenance document analysis unit 200, a work part correspondence extraction unit 300, a recommendation data storage unit 400, and a planner terminal 600. Of these components, the maintenance document analysis unit 200 and the work part correspondence extraction unit 300 are realized through the operation of a server (not shown). The business data storage unit 100 and the recommendation data storage unit 400 are storage units that constitute the server (not shown).

[0014] The business data storage unit 100 stores maintenance documents and parts history related to past failure cases. The business data storage unit 100 consists of a maintenance document table 101, a parts history table 102, a planner table 103, and a work word table 104.

[0015] The maintenance document analysis unit 200 acquires data from the business data storage unit 100, processes it, stores the results in the recommendation data storage unit 400, and transmits the results to the work part correspondence extraction unit 300. The maintenance document analysis unit 200 is composed of a work content extraction unit 201, a recovery unit extraction unit 202, and a reliability calculation unit 203. The work content extraction unit 201 extracts the necessary work content from the maintenance document database, which is a history containing documents that record the work instructed or performed to restore equipment that has malfunctioned in the past. For each maintenance document related to a single malfunction case, the work content extraction unit 201 extracts a series of inspection work, inspection results, repair work, and the contents described in those maintenance documents, all targeting a part of the equipment, as well as the restoration unit.

[0016] The work parts correspondence extraction unit 300 processes the transmitted data and stores the results in the recommendation data storage unit 400. These processes are performed periodically, such as when data is added to the business data storage unit 100. The work parts correspondence extraction unit 300 includes a recovery unit parts estimation unit 301, a recovery unit aggregation unit 302, and a recommendation matrix calculation unit 303.

[0017] The work part correspondence extraction unit 300 extracts work part correspondences by estimating which work content a part number from the history of recording part numbers that were expected to be used when the work was instructed or when the work was performed corresponds to. When information identifying a failure that has occurred in the equipment is input to the recommendation matrix calculation unit 303, it calculates a recommendation score for the necessary work or part. The recommendation matrix calculation unit 303 calculates the recommendation score by weighting it according to the degree of confidence based on the system's operating time and the amount of data.

[0018] The recovery unit part estimation unit 301 estimates and assigns part numbers that are expected to be used or used in repair work by comparing recovery units extracted from maintenance documents containing records of work performed in the same failure case with recovery units extracted from maintenance documents containing records of work performed in the same failure case, or by comparing recovery units extracted from maintenance documents in different failure cases. The recovery unit aggregation unit 302 assigns work-part correspondence relationships between work and parts included in the same recovery unit for all extracted recovery units, using work and parts as keys, and further treats work of the same type and parts of the same type as the same, and assigns work-part correspondence relationships between work and parts.

[0019] User 700 is a maintenance planner who plans and instructs the parts and work required to recover from a malfunction. User 700 operates the planner terminal 600 to understand the work and parts required to recover from a malfunction. This planner terminal 600 consists of a recovery unit recommendation unit 6010, a recommendation basis display unit 6020, and a maintenance document draft display unit 6030.

[0020] The recovery unit recommendation unit 6010 first displays the recovery unit recommendation screen 601 on the display unit. The recovery unit recommendation unit 6010 then recommends the parts corresponding to the work calculated by the recommendation matrix calculation unit 303, and the work corresponding to the parts calculated by the recommendation matrix calculation unit 303, by supplementing them with the work part correspondence relationship table 405.

[0021] The recovery unit recommendation unit 6010 of the planner terminal 600 retrieves data stored in the recommendation data storage unit 400 via the network 500 and displays the information to the user 700 on the recovery unit recommendation screen 601. The recovery unit recommendation unit 6010 displays the recommendation basis screen 602 on the display unit.

[0022] The maintenance document draft display unit 6030 displays the maintenance document draft screen 603 on the display unit. The maintenance document draft display unit 6030 shows the draft of the maintenance document created by the user by expanding the text stored in the calculated recovery unit based on the results of the work or parts recommendation, or the content of the recommendation accepted by the user. The user 700 can grasp more detailed information by operating the recovery unit recommendation screen 601 and viewing the recommendation basis screen 602 and the maintenance document draft screen 603.

[0023] In this invention, fault recovery refers to the process of inspecting one or more parts of a malfunctioning device and performing repairs and parts replacements based on the results of those inspections. Here, a series of inspections, repairs, and parts replacements for a single part is referred to as a recovery unit. To expedite fault recovery, it is crucial to complete all necessary recovery units during a single on-site visit. Therefore, it is necessary to determine the required recovery units in advance, plan the work content and parts to be carried, and then travel to the site.

[0024] [Business data storage unit 100] Figure 2 shows an example of the data structure of the preservation document table 101. Each row in the maintenance document table 101 stores in the work details column 1017 the details of the work instructed or reported for a particular failure indicated in the failure case ID column 1011. The maintenance document table 101 consists of the maintenance document ID column 1010, failure case ID column 1011, occurrence date column 1012, planner ID column 1013, failure type column 1014, instruction / report category column 1015, business trip number column 1016, and work details column 1017.

[0025] The Maintenance Document ID field 1010 stores the identifier of the maintenance document. The Failure Case ID field 1011 stores the failure case ID for the maintenance work. The Occurrence Date field 1012 stores the date and time the failure occurred. The Planner ID field 1013 stores the identifier of the planner for the maintenance work. The Failure Type field 1014 stores information about the type of failure. The Instruction / Report Classification field 1015 stores whether the line is an instruction or a report. The Business Trip Number field 1016 lists the business trip numbers sequentially. The Work Details field 1017 lists the work details in bullet points.

[0026] In this case, multiple rows with the same failure case ID stored in the failure case ID field 1011 may have different business trip numbers stored in the business trip number field 1016. This is because multiple business trips may occur for a single failure.

[0027] Furthermore, instructions and reports may be stored in the instruction / report category field 1015 for multiple rows where the failure case ID stored in the failure case ID field 1011 is the same and the travel number stored in the travel number field 1016 is the same. This is because, for each travel related to a single failure, instructions for the work content are created before the travel, and a report on the work content is created after the travel.

[0028] Figure 3 shows an example of the data structure of the parts history table 102. The parts history table 102 consists of a failure case ID column 1020, a parts number column 1021, a parts name column 1022, a quantity column 1023, and a carrying / usage classification column 1024. Each row in the parts history table 102 stores the part number in the parts number column 1021 and the quantity of the parts in the quantity column 1023 for the failure to which a failure case ID was assigned. The carrying / usage classification column 1024 stores classification information indicating whether the part was merely carried or actually used.

[0029] The part name field 1022 stores a common name assigned to part numbers of the same type. For example, for parts used to replace a part of a piece of equipment, the same part name field 1022 is used for parts that have different part number fields 1021 depending on the model of the equipment.

[0030] Figure 4 shows an example of the data structure of the working word table 104. The work terminology table 104 is a table that defines important terms in the work content column 1017 of the maintenance document table 101. The work terminology table 104 consists of a word column 1030 that stores the terms and a label column 1031 that stores the type of the terms.

[0031] The label field 1031 stores "target," which indicates the part or component of the equipment; "inspection," which indicates the content of the inspection performed on that part or component; and "repair," which indicates the content of the repair or replacement. The work word table 104 is used in the processing of the work content extraction unit 201.

[0032] Figure 5 shows an example of the data structure of the working word table 104. Each row in the work term table 104 stores attribute information of the planner, as indicated in the planner ID column 1040. The work term table 104 consists of the planner ID column 1040, the job title column 1041, the year of joining the company column 1042, and the repair completion rate for the day column 1043. Furthermore, the number of years of experience can be easily calculated from the difference between the year of joining the company and the current fiscal year. The "Same-Day Repair Completion Rate" column 1043 stores the percentage of repairs completed on the same day within a range such as the most recent year. The percentage of repairs completed on the same day is calculated by determining the percentage of cases where the "On-Site Number" column 1016 for each failure case ID column 1011 in the maintenance document table 101 related to the planner ID stored in the planner ID column 1040 is 1, within a range such as the most recent year.

[0033] [Recommendation data storage unit 400] Figure 6 shows an example of the configuration of the work content table 401. The work details table 401 includes the following columns: maintenance document ID column 4010, work number column 4011, keyword column 4012, original document column 4013, result column 4014, work type column 4015, and similar work ID column 4016.

[0034] The Maintenance Document ID field 4010 stores the identifier related to the maintenance document. The Work Number field 4011 stores the number related to the maintenance work. The Keyword field 4012 lists keywords extracted from the original text of the maintenance document. The Original Text field 4013 contains the portion of the original text of the maintenance document related to keyword extraction. The Result field 4014 stores whether the inspection result from the maintenance work was successful or not. The Work Type field 4015 stores whether it is an inspection work or a repair work. The Similar Work ID field 4016 stores the identifiers of other works similar to this maintenance work.

[0035] The work content extraction unit 201 identifies the natural document containing the work content from the work content column 1017 of the maintenance document table 101, based on the identifier of the maintenance document stored in the maintenance document ID column 4010. By processing the natural document containing the work content, the work content extraction unit 201 extracts identifiers of other similar work, keywords described in each natural document, and the results of that work. The work content extraction unit 201 further stores the identifiers of other similar work in the similar work ID column 4016, the keywords described in each natural document in the keyword column 4012, and the results of that work in the results column 4014.

[0036] The task number field 4011 is the order in which the keywords in the keyword field 4012 appear in the natural language document describing the task stored in the task details field 1017. The task type field 4015 is the estimated result of the task type, and either "inspection" or "repair" is stored there. The similar task ID field 4016 is assigned the same identifier to tasks whose keyword fields 4012 are similar.

[0037] Figure 7 shows an example of the configuration of the recovery unit table 402. The recovery unit table 402 consists of a recovery unit ID column 4020, a failure case ID column 4021, a work sequence column 4022, a similar work ID column 4023, a part number column 4024, a part name column 4025, a reliability column 4026, and an abnormality presence / absence column 4027.

[0038] Each row in the recovery unit table 402 stores a recovery unit that includes a series of inspection, repair, and parts replacement operations targeting a single part, as well as similar operations, extracted by the recovery unit extraction unit 202. Here, the recovery unit ID column 4020 stores a unique identifier for all failure cases. The similar operation ID column 4023 stores identifiers for operations similar to this series of inspection, repair, and parts replacement operations targeting this single part.

[0039] The work sequence column 4022 stores the work sequence. The recovery unit extraction unit 202 stores the identifiers obtained by renumbering the similar work ID column 4016 of the work content table 401 in ascending order using the failure case identifier stored in the failure case ID column 4021, as the work number column 4011 of the corresponding row.

[0040] The part number field 4024 stores the part number for this recovery unit. The part name field 4025 stores the name of the part for this recovery unit.

[0041] The abnormality status column 4027 is a column that stores information indicating whether or not there is an abnormality in the recovery unit. The recovery unit extraction unit 202 refers to the result column 4014 of the corresponding row in the work content table 401 for the recovery unit and stores NO in the abnormality status column 4027 if there is at least one NG indicating the discovery of an abnormality, or YES if there are no NGs indicating the discovery of an abnormality.

[0042] The reliability column 4026 stores the reliability of the information in the recovery unit. The reliability calculation unit 203 calculates the reliability of the maintenance document related to the failure case ID column 4021 and stores the value in the reliability column 4026.

[0043] Figure 8 shows an example of the configuration of the work parts correspondence table 405. The work part correspondence table 405 consists of a similar work ID column 4030, a part name column 4031, and a recovery unit ID column 4032.

[0044] The Similar Work ID field 4030 stores the identifier of a work similar to this repair work. The Part Name field 4031 stores the names of the parts required for that work. The Recovery Unit ID field 4032 stores the identifier of the recovery unit that includes this set.

[0045] Figure 9A shows an example of the configuration of the fault type component matrix 407. The failure type component matrix 407 has a component number column 4071 in the rows and a failure type column 4072 in the columns. The failure type component matrix 407 stores a value that represents the probability that the component in component number column 4071 was brought to the site or used for repair in a failure type column 4072, weighted by the confidence level of failure cases in which this component was used.

[0046] Figure 9B shows an example of the configuration of the failure type work matrix 406. The failure type work matrix 406 stores the probability that a similar work will be instructed or planned, with the rows being the similar work ID column 4061 and the columns being the failure type column 4062. The failure type work matrix 406 stores a value that is a weighted number representing the probability that the work in the similar work ID column 4061 will be used for repair in the failure type column 4062, weighted by the confidence level of the failure case in which this work was used. The failure type part matrix 407 and the failure type work matrix 406 are intermediate data used in the processing of the recommendation matrix calculation unit 303.

[0047] Figure 10 shows an example of the configuration of the component recommendation matrix 404. The part recommendation matrix 404 takes part number columns 4041 as rows and failure type columns 4042 as columns, and stores the recommendation degree for carrying the part in part number column 4041 in the failure type column 4042, which is the result of processing by the recommendation matrix calculation unit 303. The task recommendation matrix 403 takes similar task IDs as rows and similarly stores the recommendation degree for performing the task. An example of the configuration of the task recommendation matrix 403 is omitted.

[0048] [Preservation Document Analysis Unit 200] The maintenance document analysis unit 200 acquires data from the business data storage unit 100, processes it using the work content extraction unit 201, the recovery unit extraction unit 202, and the reliability calculation unit 203, inputs the results into the work part correspondence extraction unit 300 for processing, and stores the results in the recommendation data storage unit 400. The maintenance document analysis unit 200 performs its processing at a frequency such as once a day, or when new data is added to the business data storage unit 100. The work content extraction unit 201 retrieves the preservation document table 101 and the work word table 104 from the business data storage unit 100, extracts the work information contained in the work content column 1017, and stores it in the work content table 401.

[0049] Figure 11 is a flowchart of the processing performed by the work content extraction unit 201. First, the work content extraction unit 201 obtains the maintenance document table 101 and its word column 1030 (step S10). The work content extraction unit 201 obtains one unprocessed row from the maintenance document table 101 (step S11), and extracts all the terms contained in the word column 1030 of the work word table 104 for the work content column 1017 of the obtained row (step S12). Then, the work content extraction unit 201 divides the extracted terms by punctuation marks and delimiters such as "·" and "1." (step S13).

[0050] Furthermore, the work content extraction unit 201 divides the text into pairs of object labels, such as "target" which indicates the object, and predicate labels, such as "inspection" and "repair" which indicate the predicate (step S14). The work content extraction unit 201 stores the list of divided words in the keyword field 4012, stores the type of work corresponding to the type of divided predicate label in the work type field 4015, and stores the sentences from which the divided words have been extracted in the work content field 1017 to the original text field 4013, updating the work content table 401. The work content extraction unit 201 stores the sequential number of the sentence in the order in which it appeared in the work content in the work sequence field 4022 (step S15).

[0051] The work content extraction unit 201 stores the results extracted by pattern matching from the text stored in the original text field in the results field, and the measured values ​​extracted by pattern matching in the measured value field, in the work content table 401 (step S16). The work content extraction unit 201 stores the text from which the keyword field 4012 was extracted in the original text field 4013, and extracts inspection results such as "normal" or "OK" by pattern matching and stores them in the results field 4014. At this time, "NG" is stored for negative results and "OK" for positive results. The work type is "repair" if the term in the keyword field 4012 contains the term labeled "repair", and "inspection" otherwise.

[0052] Then, in step S17, the work content extraction unit 201 determines whether all rows of the preservation document table 101 have been processed. If any row is unprocessed (No), it returns to step S11. In step S17, if the work content extraction unit 201 has processed all rows of the preservation document table 101 (Yes), it proceeds to step S18.

[0053] In step S18, the work content extraction unit 201 retrieves word pairs from the keyword column 4012 of the work content table 401 and calculates feature quantities such as word2vec for each row. Then, the work content extraction unit 201 assigns the same similar work ID to rows where the distance of the feature quantities is less than or equal to a threshold and stores them in the work content table 401 (step S19), at which point the process shown in Figure 11 is completed.

[0054] Figure 12 is a flowchart of the processing performed by the recovery unit extraction unit 202. As described above, in this invention, a recovery unit refers to a series of inspections, repairs, and parts replacements performed on a single part of equipment that has experienced a malfunction, in order to restore it to normal operation.

[0055] The recovery unit extraction unit 202 obtains word pairs from the keyword column 4012 of the work content table 401 and calculates feature quantities such as word2vec for each row (step S30). Specifically, the recovery unit extraction unit 202 calculates known feature quantities such as word2vec for the keyword column 4012 and considers them similar if the distance is below a threshold. The recovery unit extraction unit 202 first calculates the contents of the similar work ID column 4016 of the work content table 401. The similar work ID column 4016 stores the same identifier for work whose contents in the keyword column 4012 are similar and whose work type is the same. The recovery unit extraction unit 202 then retrieves rows of unprocessed failure case IDs from the work content table 401 (step S31), and repeats the process of extracting a series of inspection and repair tasks from steps S32 to S38.

[0056] In step S32, the recovery unit extraction unit 202 stores the information from the similar work ID field 4016, the work sequence, and the original text field 4013 for rows where the work numbers in the work number field 4011 are consecutive and the distance of the calculated keyword feature quantities is below a threshold, into the respective fields of the recovery unit table 402 with the same names. It then stores the result field 4014 into the abnormality presence / absence field 4027 of the recovery unit table 402 and assigns the same recovery unit ID to the recovery unit ID field 4020.

[0057] In step S33, the recovery unit extraction unit 202 retrieves a row containing one unprocessed recovery unit ID from the recovery unit table 402. Then, the recovery unit extraction unit 202 determines whether or not there is an "NG" in the "Abnormality" column 4027 of the retrieved row (step S34). If there is an "NG" in the "Abnormality" column 4027 of the retrieved row (Yes), the recovery unit extraction unit 202 replaces all of them with "Yes" (step S36). If all of the "Abnormality" columns 4027 of the retrieved row are "YES" (No), the recovery unit extraction unit 202 replaces all of them with "No" (step S35).

[0058] In step S37, the recovery unit extraction unit 202 determines whether all recovery unit IDs have been processed. If any recovery unit ID is unprocessed (No), the recovery unit extraction unit 202 returns to step S33. If all recovery unit IDs have been processed (Yes), the recovery unit extraction unit 202 proceeds to step S38.

[0059] In step S38, the recovery unit extraction unit 202 determines whether all rows in the work content table 401 have been processed. If any row in the work content table 401 is unprocessed (No), the recovery unit extraction unit 202 returns to step S32. If all rows in the work content table 401 have been processed (Yes), the recovery unit extraction unit 202 terminates the process shown in Figure 12. The part number column 4024, the part name column 4025, and the reliability column 4026 are calculated by subsequent processing.

[0060] Specifically, the recovery unit extraction unit 202 considers, for each recovery unit ID column 4020, tasks where the task number column 4011 is consecutive and the aforementioned keyword column 4012 is similar as a series of inspection and repair tasks, without considering the differences in the task type column 4015. The recovery unit extraction unit 202 stores the same identifier in the recovery unit ID column 4020 of the recovery unit table 402 for tasks where the task number column 4011 is consecutive and the aforementioned keyword column 4012 is similar. In addition, the recovery unit extraction unit 202 assigns and stores the task sequence column 4022 starting from 1 for each recovery unit ID column 4020 according to the task number column 4011 of the task content table 401. The recovery unit extraction unit 202, for each recovery unit ID column 4020 in the work content table 401, sets all abnormality column 4027 to "YES" if it contains NG, and sets all abnormality column 4027 to "NO" otherwise.

[0061] Figure 13 is a flowchart of the processing performed by the confidence calculation unit 203. The reliability calculation unit 203 obtains one unprocessed failure case ID from the recovery unit table 402 (step S40). The reliability calculation unit 203 obtains the business trip number of the acquired failure case ID from the business trip number column 1016 of the maintenance document table 101, calculates the maximum value as the number of business trips, and obtains the instruction report from the instruction report category column 1015 (step S41). The reliability calculation unit 203 obtains the planner ID of the acquired failure case ID from the planner ID column 1013 of the maintenance document table 101 (step S42), obtains the job title column 1041, the year of joining the company column 1042, and the repair completion rate column 1043 of the acquired planner ID from the planner table 103, and calculates the reliability based on the acquired parameters and the calculated number of business trips (step S43). The reliability calculation unit 203 stores the reliability calculated for the acquired failure case ID in the reliability column 4026 of the recovery unit table 402 (step S44). In step S45, the reliability calculation unit 203 determines whether all rows in the work content table 401 have been processed. If there are any unprocessed rows in the work content table 401 (No), the reliability calculation unit 203 returns to step S40. If all rows in the work content table 401 have been processed (Yes), the reliability calculation unit 203 terminates the process shown in Figure 13.

[0062] In other words, the reliability calculation unit 203 retrieves the maintenance document table 101, the planner table 103, and the recovery unit table 402, calculates the reliability for the contents of each maintenance document ID, and stores it in the reliability column 4026 of the recovery unit table 402. Here, the reliability includes the attributes of the target failure case of the maintenance document, the attributes of the planner who created the maintenance document, and the attributes of the document type. Specifically, as an attribute of the target failure case of the maintenance document, the number of business trips required to recover from that failure case can be used, and the fewer the number of business trips, the higher the reliability. The number of business trips for the target failure case is calculated by calculating the maximum value of the business trip number column 1016 for each failure case ID column 1011 in the maintenance document table 101.

[0063] As attributes of the planner who creates the maintenance document, the higher the position in the position column 1041 of the planner table 103, the longer the number of years of experience in the year of joining the company column 1042, and the higher the same-day repair completion rate (the rate at which a malfunction is resolved in a single business trip), the more reliable the planner is considered to be.

[0064] As an attribute of the document type, if the instruction / report category column 1015 in the maintenance document table 101 for the relevant failure case is "instruction," which is assigned to documents created before the business trip, the reliability is low, and if it is "report," which is assigned to documents created after the business trip, the reliability is high.

[0065] As an example of a specific formula for calculating confidence using these attributes, the following formula (1) can be used.

number

[0066] [Work parts correspondence extraction unit 300] The work parts correspondence extraction unit 300 acquires the processing results from the maintenance document analysis unit 200 and the data from the business data storage unit 100, performs processing by the recovery unit parts estimation unit 301, the recovery unit aggregation unit 302, and the recommendation matrix calculation unit 303, and stores the results in the recommendation data storage unit 400.

[0067] Figures 14A to 14C are flowcharts of the processing performed by the recovery unit component estimation unit 301. The recovery unit part estimation unit 301 estimates and stores the part number column 4024 and the part name column 4025 of the recovery unit table 402. The recovery unit component estimation unit 301 obtains one unprocessed failure case ID from the recovery unit table 402 (step S50), and obtains the recovery unit ID and whether or not there is an abnormality for the obtained failure case ID from the recovery unit table 402 (step S51).

[0068] The recovery unit component estimation unit 301 obtains the instruction report category of the acquired failure case ID from the maintenance document table 101 (step S52). The recovery unit component estimation unit 301 obtains the component number, component name, and carrying / usage classification of the acquired failure case ID from the component history table 102 (step S53).

[0069] In other words, the recovery unit part estimation unit 301 retrieves all part number fields 1021 and carrying / usage category fields 1024 related to one failure case ID field 1020 from the parts history table 102, and similarly retrieves all recovery unit ID fields 4020 and abnormality / absence fields 4027 that match the failure case ID field 4021 from the recovery unit table 402, and all instruction / report category fields 1015 that match the failure case ID field 1011 from the maintenance document table 101. Then, the recovery unit part estimation unit 301 repeats the following steps S54 to S60 for all part numbers.

[0070] The recovery unit part estimation unit 301 obtains one unprocessed part number (step S54). The recovery unit component estimation unit 301 obtains all the portable usage categories for the acquired part numbers from the portable usage category column 1024 related to the acquired failure case ID (step S55).

[0071] In step S56, the recovery unit part estimation unit 301 determines whether the carrying and usage category related to the part number is "carrying and usage". If the carrying and usage category is "carrying and usage" (Yes), the unit proceeds to step S57. If the recovery unit part estimation unit 301 determines that the carrying and usage category related to the part number is not "carrying and usage" (No), the unit proceeds to step S58.

[0072] In step S57, the recovery unit part estimation unit 301 stores the recovery unit ID, which has obtained the part number and part name, in the recovery unit table 402, where the instruction report category is instruction and report, and the obtained abnormality status is YES.

[0073] In step S58, the recovery unit part estimation unit 301 determines whether the portable use category is portable only. If the portable use category is portable only (Yes), the unit proceeds to step S59. If the portable use category related to the part number is not portable only (No), the unit proceeds to step S60.

[0074] In step S59, the recovery unit part estimation unit 301 stores the recovery unit ID, which has obtained the part number and part name, in the recovery unit table 402, where the instruction report category is "instruction" and the obtained abnormality status is "NO".

[0075] In step S60, the recovery unit part estimation unit 301 determines whether all part numbers have been processed. If there are any unprocessed part numbers (No), the recovery unit part estimation unit 301 returns to step S54. If all part numbers have been processed (Yes), the recovery unit part estimation unit 301 proceeds to step S61.

[0076] The recovery unit part estimation unit 301 determines for each acquired part number field 1021 whether the carry / use category field 1024 is both carry and use, or carry only. If the carry / use category field 1024 is both carry and use, it is assumed that the part was carried as instructed, and that the part was used for repair because an abnormality was found during the instructed inspection. The part is then stored in the part number field 4024 and part name field 4025 of the recovery unit ID for which the acquired instruction / report category field 1015 is instruction and report, and the acquired abnormality presence / absence field 4027 is YES.

[0077] On the one hand, when the Carry-Use Classification field 1024 is "Carry-Only", the Recovery Unit Part Estimation Unit 301 assumes that the instructed part was carried, but was not used because no abnormalities were found during inspection. In this case, the acquired instruction report classification field 1015 is "Instruction-Only", and the acquired abnormality presence / absence field 4027 is "NO", and the part is stored in the part number field 4024 and part name field 4025 of the Recovery Unit ID field 4020.

[0078] In step S61, the recovery unit component estimation unit 301 determines whether all failure case IDs have been processed. If there are any unprocessed failure case IDs (No), the recovery unit component estimation unit 301 returns to step S50. If all failure case IDs have been processed (Yes), the recovery unit component estimation unit 301 proceeds to step S62.

[0079] In step S62, the recovery unit part estimation unit 301 retrieves all work types for similar work IDs in each row of the recovery unit table 402 from the work content table 401. Then, the recovery unit part estimation unit 301 deletes all part numbers and part names for rows where the retrieved work type is inspection work (step S63).

[0080] Depending on the processing described above, there may be cases where the Similar Work ID field 4023 and the Part Name field 4025 do not have a one-to-one relationship. In such cases, it is desirable for the Recovery Unit Part Estimation Unit 301 to decompose them into a one-to-one relationship using estimation results from other failure cases. To do this decomposition, the Recovery Unit Part Estimation Unit 301 extracts those in each Recovery Unit ID field 4020 that do not have a one-to-one relationship, i.e., those in which the Part Name field 4025 contains two or more entries. The Recovery Unit Part Estimation Unit 301 then counts the number of cases where either the Similar Work ID field 4023 or the two or more Part Name fields 4025 for each Recovery Unit ID field 4020 matches the Similar Work ID field 4023 and Part Name field 4025 of all other Recovery Units that have a one-to-one relationship. The Recovery Unit Part Estimation Unit 301 deletes the remaining part numbers, keeping only the part with the highest number of matches. At this time, the Recovery Unit Part Estimation Unit 301 keeps all of the matching entries if they are the same number.

[0081] Steps S64 through S70 below are a loop that processes all pairs of recovery unit IDs and similar work IDs. The recovery unit component estimation unit 301 obtains one pair of an unprocessed recovery unit ID and a similar work ID from the recovery unit table 402 (step S64).

[0082] The recovery unit part estimation unit 301 obtains all rows and part numbers of the acquired recovery unit ID and similar work ID (step S65). In step S66, the recovery unit part estimation unit 301 determines whether there are multiple part numbers. If there is only one part number (No), the recovery unit part estimation unit 301 proceeds to step S70. If there are multiple part numbers (Yes), the recovery unit part estimation unit 301 proceeds to step S67.

[0083] In step S67, the recovery unit part estimation unit 301 obtains the part name of the acquired part number and then obtains a recovery unit ID that matches the acquired part name with a similar work ID (step S68).

[0084] The recovery unit part estimation unit 301 deletes all rows except the one with the highest number of recovery unit IDs that match the part name and similar work ID (step S69). The recovery unit component estimation unit 301 determines whether all pairs of recovery unit IDs and similar work IDs have been processed (step S70). If there are any pairs of recovery unit IDs and similar work IDs that have not been processed (No), the recovery unit component estimation unit 301 returns to step S64. If all pairs of recovery unit IDs and similar work IDs have been processed (Yes), the recovery unit component estimation unit 301 terminates the process shown in Figure 14C.

[0085] Figure 15 is a flowchart of the processing performed by the recovery unit aggregation unit 302. The recovery unit aggregation unit 302 retrieves the recovery unit table 402, generalizes and extracts the relationship between the work and parts, and stores it in the work-part correspondence relationship table 405. The recovery unit aggregation unit 302 retrieves one unprocessed part name from the recovery unit table 402 (step S110).

[0086] The recovery unit aggregation unit 302 obtains all of the acquired part names, which are recovery unit IDs (step S110). The recovery unit aggregation unit 302 repeats the following processes from steps S112 to S116 for all recovery unit IDs.

[0087] The recovery unit aggregation unit 302 obtains one unprocessed recovery unit ID (step S112) and obtains a similar work ID from the recovery unit ID (step S113).

[0088] The recovery unit aggregation unit 302 retrieves the work type of the acquired similar work ID from the work content table 401 (step S114), and stores the acquired part name and work type of the similar work ID and recovery unit ID that are repairs in the column with the same name in the work part correspondence relationship table 405 (step S115). In step S116, the recovery unit aggregation unit 302 determines whether all recovery unit IDs have been processed. If there are any unprocessed recovery unit IDs (No), the recovery unit aggregation unit 302 returns to step S112. If all recovery unit IDs have been processed (Yes), the recovery unit aggregation unit 302 proceeds to step S117.

[0089] In step S117, the recovery unit aggregation unit 302 determines whether all part names have been processed. If there are any part names that have not been processed (No), the recovery unit aggregation unit 302 returns to step S110. If all part names have been processed (Yes), the recovery unit aggregation unit 302 terminates the process shown in Figure 15.

[0090] Specifically, the recovery unit aggregation unit 302 extracts the similar work ID column 4023 and recovery unit ID column 4020 from the row in the recovery unit table 402 where the part name column 4025 is stored. The recovery unit aggregation unit 302 further extracts only those similar work ID columns 4023 where the work type column 4015 of the work content table 401 is "repair" and stores them in the work part correspondence table 405.

[0091] Figures 16A and 16B are flowcharts of the processing performed by the recommendation matrix calculation unit 303. The recommendation matrix calculation unit 303 retrieves the maintenance document table 101, the work content table 401, and the parts history table 102. For each failure in the failure type column 1014 of the maintenance document table 101, it calculates a recommendation score to recommend similar work IDs and part numbers for the work required to restore the failure. The results are then stored in the work recommendation matrix 403 and the parts recommendation matrix 404. In this embodiment, the method for calculating the parts recommendation matrix 404, which stores the parts recommendation scores, is shown first. However, in subsequent processing, the part name may be used instead of the part number.

[0092] The processing in steps S80 to S86 is a repetitive process for all types of failures. The recommendation matrix calculation unit 303 retrieves one unprocessed failure type from the maintenance document table 101 (step S80) and retrieves all failure case IDs for the retrieved failure type (step S81).

[0093] The recommendation matrix calculation unit 303 obtains the confidence level of the acquired failure case ID and the similar work ID from the confidence level column 4026 and the similar work ID column 4023 of the recovery unit table 402 (step S82). The recommendation matrix calculation unit 303 uses the acquired failure types as rows and the acquired similar task IDs as columns, calculates the probability of performing a similar task ID for that failure type, weights the calculated execution probabilities based on the acquired confidence level, and stores them in the failure type task matrix 406 (step S83).

[0094] The recommendation matrix calculation unit 303 obtains the confidence level and part number of the acquired failure case ID from the confidence level column 4026 and part number column 4024 of the recovery unit table 402 (step S84). The recommendation matrix calculation unit 303 uses the acquired failure types as rows and the acquired part numbers as columns, calculates the probability of performing a similar operation ID for that failure type, weights the calculated execution probabilities based on the acquired confidence level, and stores them in the failure type / part matrix 407 (step S85).

[0095] The recommendation matrix calculation unit 303 determines whether all fault types have been processed (step S86). If there are any fault types that have not been processed (No), the recommendation matrix calculation unit 303 returns to step S80. If all fault types have been processed (Yes), the recommendation matrix calculation unit 303 proceeds to step S87.

[0096] In step S87, the recommendation matrix calculation unit 303 applies a non-negative matrix decomposition method to the failure type work matrix 406, multiplies the calculated matrix with failure types as rows by the calculated matrix with similar work IDs as columns, and stores the result in the work recommendation matrix 403. The recommendation matrix calculation unit 303 applies a non-negative matrix decomposition method to the fault type component matrix 407, multiplies the calculated matrix with fault types as rows by the calculated matrix with component numbers as columns, and stores the result in the component recommendation matrix 404 (step S88), at which point the process shown in Figure 16B is completed.

[0097] To calculate the parts recommendation matrix 404, first, the failure type column 1014 is used as a row, and the parts number column 1021, which is carried or used in the failure case ID column 1011 of the relevant failure type, is obtained from the parts history table 102 and used as a column. The value of this column is calculated to store the rate at which the parts number is carried or used in the failure case ID column 1020 of the relevant failure type. This is used to calculate the failure type parts matrix 407. Furthermore, the reliability column 4026 of the failure case ID column 4021 is averaged for each instruction report category column 1015 and obtained from the recovery unit table 402, multiplied by an appropriate coefficient, and then multiplied as the weight of the above value. The non-negative matrix decomposition method, a known technique, is applied to the failure type parts matrix 407 calculated in this way to decompose it into a matrix with failure types as rows and a matrix with parts numbers as columns, and the result of multiplying these decomposed matrices is stored in the parts recommendation matrix 404. The values ​​obtained by referencing the failure type and part number in the part recommendation matrix 404 represent the recommendation score, which indicates the degree to which the part number is recommended for that failure type. Similar failure types tend to have higher recommendation scores for the same part number. By using this part recommendation matrix 404, it is possible to identify all the necessary sets of failure types and part numbers for recovery, even for combinations that have not been recorded in past data.

[0098] The work recommendation matrix 403 and the failure type work matrix 406 are calculated by replacing the part number in the calculation method with a similar work ID. In addition to the work recommendation matrix 403 using a similar work ID, a similar configuration is also envisioned for calculating a matrix using a recovery unit ID.

[0099] [Recovery Unit Recommendation Screen 601] Figure 17 shows an example of the recovery unit recommendation screen 601. The recovery unit recommendation unit 6010 acquires data from the business data storage unit 100 and the recommendation data storage unit 400 via the network 500, and notifies the user 700 of the information by displaying it on the recovery unit recommendation screen 601.

[0100] The failure type combo box 6011 is a combo box for the user to input the type of failure. The selection confirmation button 6012 is a button for confirming the selected results, such as the work to be performed or the parts to be carried.

[0101] The task checkbox 6013 is a button for selecting a task to perform. When checked, it means that this task will be performed. The "Carry Parts" checkbox 6014 is a button for selecting parts to carry; when checked, it means that this part will be carried. The recommendation basis display button 6015 is a button used to display the recommendation basis screen 602 related to the implemented work.

[0102] Figure 18 is a sequence diagram of the process related to the display of the recovery unit recommendation screen 601. The recovery unit recommendation unit 6010 performs the following processes sequentially each time the user 700 operates the recovery unit recommendation screen 601. When the user 700 enters the type of failure into the failure type combo box 6011, the recovery unit including the work and parts necessary to recover from that failure is displayed on the recovery unit recommendation screen 601.

[0103] In step S120, the user enters the type of failure into the failure type combo box 6011. This prompts the recovery unit recommendation unit 6010 to display the information (step S121). The recovery unit recommendation unit 6010 obtains the parts / work with the highest recommendation scores from the parts recommendation matrix 404 and the work recommendation matrix 403 (step S122). Specifically, the recovery unit recommendation unit 6010 obtains the parts recommendation matrix 404 and the work recommendation matrix 403 and refers to the rows corresponding to the input failure types.

[0104] The recovery unit recommendation unit 6010 retrieves work units of work / parts related to the recommendation candidate from the work part correspondence relationship table 405 (step S123). The recovery unit recommendation unit 6010 obtains work units by referring to the recovery unit table 402 (step S124). Then, the recovery unit recommendation unit 6010 presents candidates via the recovery unit recommendation screen 601 (step S125).

[0105] The recovery unit recommendation unit 6010 repeats the processes in steps S126 and S127 each time the user selects a part or operation (step S125) until the user presses the plan completion button. In step S126, the recovery unit recommendation unit 6010 displays a warning dialog and, if any parts have been selected, displays their weight (step S127).

[0106] The system retrieves the top few similar work IDs and part numbers for the columns with the highest recommendation scores, obtains the work / part correspondence table 405, retrieves all the recovery unit IDs 4020 corresponding to the retrieved similar work IDs, and retrieves all the recovery unit IDs 4025 corresponding to the retrieved part names 4025. The contents of the recovery unit IDs 4020 are then retrieved from the recovery unit table 402 and displayed on the recovery unit recommendation screen 601.

[0107] The recovery unit recommendation unit 6010 retrieves the keyword field 4012 and work type field 4015 of the work for each acquired recovery unit from the work content table 401, and displays the keyword field 4012 and work type field 4015 in the work check box 6013 in ascending order of the work sequence field 4022 for each recovery unit ID field 4020. At this time, the recovery unit recommendation unit 6010 groups together recovery units where the majority of the work in the work type field 4015 is "inspection" and displays them after removing duplicates.

[0108] Here, for tasks in the recovery unit table 402 with a similar task ID column 4023 that has a part name column 4025, the recovery unit recommendation unit 6010 displays the part name column 4025 and part number column 4024 of the related part in the carried part checkbox 6014, and further retrieves and displays the median value of the number of times the part was used in that failure type column 1023 from the part history table 102. In addition, for each task in the executed task checkbox 6013, the recommendation score value of the similar task ID column 4023 is displayed. However, if there are related parts, the average value of the recommendation score of the part number column 4024 and the related part is calculated and displayed. A higher recommendation score indicates a higher probability of being necessary for failure recovery.

[0109] The parts and tasks placed in the task checkbox 6013 and the parts to carry checkbox 6014 can be switched by the user 700 to either perform the task or carry the parts. A warning is displayed if only one of the related tasks or parts is selected.

[0110] If user 700 wishes to check the basis for the recommendation of each recovery unit, they press the recommendation basis display button 6015 and are taken to the recommendation basis screen 602. Also, once user 700 has finished considering all tasks and parts selections, they press the selection confirmation button 6012 and are taken to the maintenance document draft screen 603.

[0111] [Recommendation rationale screen 602] Figure 19 shows an example of the screen layout of the recommendation basis screen 602. The recommendation basis screen 602 retrieves data from the business data storage unit 100 and the recommendation data storage unit 400 via the network 500 and displays past failure cases that serve as the basis for the recommendation on the recovery unit recommendation screen 601.

[0112] Title label 6021 indicates the title of the report document for this maintenance work. Case selection button 6022 is used to view and select other case studies besides the one currently displayed.

[0113] The reliability breakdown label 6023 is a label that displays a breakdown of each parameter in the reliability column 4026 for the acquired failure case ID, similar to the processing by the reliability calculation unit 203. The recovery unit combo box 6024 is a combo box that allows the user to select from a list of keywords for tasks where the failure type of the recovery unit performed in the acquired failure case ID is inspection. The recovery unit document display area 6025 is an area that displays the original document and part number corresponding to the recovery unit ID.

[0114] Figure 20 is a flowchart of the processing of the recommendation basis display unit 6020. The recommendation basis display unit 6020 sequentially executes the following processes each time the user 700 presses the recommendation basis display button 6015 on the recovery unit recommendation screen 601. The recommendation basis display unit 6020 obtains the failure case ID of the failure case displayed by the user (step S90). The recommendation basis display unit 6020 obtains the confidence level of the acquired failure case ID from the recovery unit table 402 (step S91). The recommendation basis display unit 6020 obtains the business trip number of the acquired failure case ID from the maintenance document table 101, calculates the maximum value of these numbers as the number of business trips, and obtains the instruction report from the instruction report category column 1015 (step S92). The recommendation basis display unit 6020 obtains the planner ID of the acquired failure case ID from the maintenance document table 101 (step S93). The recommendation basis display unit 6020 retrieves the job title, year of joining the company, and repair completion rate for the day of the acquired planner ID from the planner table 103 (step S94). The recommendation basis display unit 6020 displays the acquired confidence level and its breakdown on the confidence level breakdown label 6023 (step S95). The recommendation basis display unit 6020 obtains the recovery unit ID of the acquired failure case ID from the recovery unit table 402 (step S96). The recommendation basis display unit 6020 retrieves the original document of the acquired recovery unit ID from the work content table 401 (step S97). The recommendation basis display unit 6020 displays the acquired recovery unit and original document in the recovery unit document display area 6025 (step S98), and then terminates the process shown in Figure 20.

[0115] The recommendation basis display unit 6020 calculates the reliability of a maintenance document based on attribute information including the document creator's years of experience, position, same-day repair completion rate, type of maintenance document, and the number of business trips required to restore the malfunction. It then displays the maintenance document and its reliability to the user as the basis information for recommending this work or part.

[0116] The recommendation basis screen 602 is given the recovery unit ID and failure type of the recovery unit for which the user has selected the basis to check. First, the reliability field 4026 for each failure case ID field 4021 that performed the recovery unit in the recovery unit ID field 4020 is obtained from the recovery unit table 402, and the failure case ID of the failure case that has the highest obtained reliability and matches the given failure type is obtained to be displayed as the basis case. However, the user 700 can specify the failure case ID to obtain by pressing the case selection button 6022. For the obtained failure case ID, the breakdown for each parameter in the reliability field 4026 is displayed in the reliability breakdown label 6023, similar to the processing of the reliability calculation unit 203. Furthermore, the original text field 4013 and part number field 4024 of the recovery unit ID field 4020 are obtained from the recovery unit table 402 and displayed in the recovery unit text display area 6025. Furthermore, the recovery unit combo box 6024 displays a list of keywords for tasks where the failure type of the recovery unit performed in the acquired failure case ID is inspection, and the user 700 can select one to switch the recovery unit displayed on the recovery unit recommendation screen 601.

[0117] [Preservation Document Draft Screen 603] Figure 21 shows an example of the preservation document draft screen 603. The maintenance document draft display unit 6030 acquires data from the business data storage unit 100 and the recommendation data storage unit 400 via the network 500, creates a maintenance document draft including the work and parts selected by the user 700 on the recovery unit recommendation screen 601, and displays it on the maintenance document draft screen 603.

[0118] Preview button 6031 is a button for previewing a draft of a work instruction document in natural language. The recovery unit number label 6032 is a label that indicates the number of the recovery unit. The work keyword label 6033 is a label that indicates keywords related to the work. The work number text box 6034 is a text box that displays the work number and consists of the major number text box 6038 and the minor number text box 6039, which will be described later. The major number text box 6038 displays the major number, which is the integer part of the work number. The minor number text box 6039 displays the minor number, which is the decimal part of the work number. The work description text box 6035 is a text box for entering the work description text. The portable parts combo box 6036 is a combo box for inputting portable parts. The support example display button 6037 is a button used to display support examples.

[0119] Figure 22 is a flowchart of the preservation document drafting process. The preservation document draft display unit 6030 sequentially executes the following processes each time the user 700 presses the selection confirmation button 6012 on the recovery unit recommendation screen 601. The maintenance document draft display unit 6030 obtains the recovery unit ID, similar work ID, and part name selected by the user (step S100).

[0120] The maintenance document draft display unit 6030 retrieves the work numbers of similar work IDs from the work content table 401 for each recovery unit ID (step S101). The maintenance document draft display unit 6030 then specifies the display order of the recovery units in ascending order of the average value of the retrieved work numbers and displays them in the work number major number text box 6038 and the recovery unit number label 6032 (step S102).

[0121] The maintenance document draft display unit 6030 retrieves the work sequence for each recovery unit from the recovery unit table 402, specifies the display order of similar work IDs in ascending order of their average values, and displays them in the sub-number text box 6039 (step S103). The maintenance document draft display unit 6030 retrieves the keyword of the similar work ID for the relevant work number from the recovery unit table 402 and displays it in the work keyword label 6033 (step S104).

[0122] The maintenance document draft display unit 6030 retrieves the original document and work type of similar work IDs for the work number from the work content table 401, displays the original document in the work content text box 6035, and if the work type is repair, retrieves whether there is an abnormality for the recovery unit ID from the recovery unit table 402, and displays the most frequent value of the retrieved abnormality status as the repair work execution condition in the work content text box 6035 (step S105).

[0123] The maintenance document draft display unit 6030 retrieves the part number of the part name obtained for the recovery unit ID of the work number from the recovery unit table 402 and displays it in the portable parts combo box 6036 (step S106), at which point it terminates the process shown in Figure 22.

[0124] The maintenance document draft screen 603 displays the recovery unit ID field 4020, the similar work ID field 4023, and the part name field 4025 selected by the user. First, the maintenance document draft display unit 6030 calculates the display order of the recovery units.

[0125] Specifically, the maintenance document draft display unit 6030 calculates the average reliability score of failure cases in which a recovery unit was performed for each recovery unit grouped together in the processing of the recovery unit recommendation screen 601. The maintenance document draft display unit 6030 then prioritizes groups of recovery units with a high average reliability score and whose failure type matches the failure type entered by the user 700 on the recovery unit recommendation screen 601, and determines the display order. The maintenance document draft display unit 6030 displays natural number values ​​in the recovery unit number label 6032 and the large number text box 6038 of the work number text box 6034 in ascending order of the display order.

[0126] Furthermore, the maintenance document draft display unit 6030 retrieves the similar work ID column 4023 for each group of recovery units from the recovery unit table 402, and displays the natural number values ​​in ascending order of the average value in the subnumber text box 6039 of the work number text box 6034.

[0127] The work keyword label 6033 and the work content text box 6035 display the keyword field 4012 and original text field 4013 of the similar work ID field 4023 of the recovery unit, obtained from the work content table 401. Furthermore, if the work type field 4015 of the similar work ID is "repair," the mode of abnormality in the recovery unit is calculated, and this value is displayed in the work content text box 6035 as "Repair work performed" in the format "Result: NG." The portable parts combo box 6036 displays all the part number fields 4024 for the recovery unit group related to the part name selected by the user 700, obtained from the recovery unit table 402. However, if there are multiple candidate part numbers, they are displayed in a format that allows the user to select from the candidates.

[0128] The recovery unit number label 6032 is equipped with a evidence case display button 6037 that the user 700 can press to check the failure cases of that recovery unit. When pressed, the system transitions to the recommendation evidence screen 602. The user 700 can freely edit or select from the options displayed in the work number text box 6034, the work content text box 6035, and the portable parts combo box 6036. Once the user 700 has finished editing or selecting for all items, they press the preview button 6031 to finish the maintenance document drafting process.

[0129] (modified version) The present invention is not limited to the embodiments described above, and includes various modifications. For example, the embodiments described above are described in detail to make the present invention easier to understand, and are not necessarily limited to those having all the configurations described. It is possible to replace parts of the configuration of one embodiment with the configuration of another embodiment, and it is also possible to add configurations from other embodiments to the configuration of one embodiment. Furthermore, it is possible to add, delete, or replace parts of the configuration of each embodiment with other configurations.

[0130] Each of the above configurations, functions, processing units, and processing means may be implemented in part or in whole by hardware, such as an integrated circuit. Each of the above configurations and functions may also be implemented in software by a processor interpreting and executing a program that implements each function. Information such as programs, tables, and files that implement each function can be stored in a recording device such as memory, a hard disk, or an SSD (Solid State Drive), or on a recording medium such as a flash memory card or a DVD (Digital Versatile Disk).

[0131] In each embodiment, the control lines and information lines shown are those deemed necessary for explanation and do not necessarily represent all control lines and information lines in the actual product. In practice, it can be assumed that almost all components are interconnected. [Explanation of Symbols]

[0132] 1. Fault Recovery Support System 100 Business Data Storage Unit 101 Conservation Document Table 102 Parts History Table 103 Planner Table 104 Working Vocabulary Table 1010 Preservation Document ID field 1011 Failure case ID column 1012 Date of occurrence column 1013 Planner ID field 1014 Failure type column 1015 Instruction report classification column 1016 Business Trip Number Field 1017 Work Details Section 1020 Failure case ID field 1021 Part Number Field 1022 Part Name Field 1023 Quantity column 1024 Portable Use Category 1030 Word list 1031 Label field 1040 Planner ID field 1041 Position field 1042 Year of joining the company 1043 Same-day repair completion rate column 200 Preservation Document Analysis Department 201 Work content extraction part 202 Recovery Unit Extraction Unit 203 Confidence Calculation Unit 300 Work Parts Correspondence Extraction Unit 301 Recovery Unit Component Estimation Unit 302 Recovery Unit Aggregation Section 303 Recommendation matrix calculation unit 400 Recommendation Data Storage Unit 401 Work Content Table 402 Recovery Unit Table 403 Work recommendation matrix 404 Component Recommendation Queue 405 Work Parts Correspondence Table 406 Failure type work matrix 407 Failure Type Component Matrix 4010 Preservation Document ID field 4011 Work Number Column 4012 Keyword field 4013 Original text field 4014 Results section 4015 Work Type Column 4016 Similar work ID field 4020 Recovery Unit ID field 4021 Failure case ID field 4022 Work order field 4023 Similar work ID field 4024 Part Number Field 4025 Part Name Field 4026 Confidence level column 4027 Abnormality column 4030 Similar work ID field 4031 Part Name Field 4032 Recovery Unit ID field 4041 Part number field 4042 Failure type column 4061 Similar work ID field 4062 Failure type column 4071 Part number field 4072 Failure type column 500 Networks 600 Planner Terminals 601 Recovery Unit Recommendation Screen 6010 Restoration Unit Recommendation Department 6011 Failure Type: Combo Box 6012 Confirm Selection Button 6013 Checklist for tasks to be performed 6014 Carry-on parts checkbox 6015 Recommendation Reason Display Button 602 Recommendation Basis Screen 6020 Recommendation Basis Display Section 6021 Title Label 6022 Case Selection Button 6023 Confidence breakdown label 6024 Recovery Unit Combo Box 6025 Recovery Unit Document Display Area 603 Preservation Document Draft Screen 6030 Preservation Document Draft Display Section 6031 Preview button 6032 Recovery Unit Number Label 6033 Work Keyword Label 6034 Work Number Text Box 6035 Work Description Text Box 6036 Portable Parts Combo Box 6037 Show evidence example button 6038 Large Number Text Box 6039 Small number text box 700 users

Claims

1. A fault recovery support system that recommends the necessary work procedures and part numbers when a malfunction occurs in equipment, A work content extraction unit extracts keywords for work content from maintenance documents, which are a history including natural documents that record the work instructed or performed to restore the equipment that has malfunctioned in the past, based on a work word table, and assigns similar work identifiers for work similar to the said work based on the feature quantities of the keywords. A work part correspondence extraction unit extracts work part correspondences by estimating which work content a part number from a history that records the part number intended to be used when the work was instructed or the part number used when the work was performed corresponds to, and When information identifying a malfunction in the aforementioned equipment is input, a recommendation matrix calculation unit calculates a similar work identifier for the necessary work and a recommendation score for the parts, A recovery unit recommendation unit recommends parts corresponding to similar work identifiers calculated by the recommendation matrix calculation unit, and work corresponding to parts calculated by the recommendation matrix calculation unit, by supplementing them using the work part correspondence relationship. A fault recovery support system characterized by comprising the following features.

2. A fault recovery support system that recommends the necessary work procedures and part numbers when a malfunction occurs in equipment, A work content extraction unit extracts necessary work content from maintenance documents, which are a history including documents recording the work instructed or performed to restore the equipment in the past, A work part correspondence extraction unit extracts work part correspondences by estimating which work content a part number from a history that records the part number intended to be used when the work was instructed or the part number used when the work was performed corresponds to, and When information identifying a malfunction in the aforementioned equipment is input, a recommendation matrix calculation unit calculates a recommendation score for the necessary work or parts. A recovery unit recommendation unit recommends the parts corresponding to the work calculated by the recommendation matrix calculation unit, and the work corresponding to the parts calculated by the recommendation matrix calculation unit, by supplementing them using the work part correspondence relationship. A recommendation basis display unit calculates the reliability of the aforementioned maintenance document based on attribute information including the document creator's years of experience, position, same-day repair completion rate, type of maintenance document, and the number of business trips required to restore the said malfunction, and shows the maintenance document and its reliability to the user as the basis information for recommending the said work or parts, A fault recovery support system characterized by comprising the following features.

3. The work content extraction unit extracts, for each maintenance document relating to a single failure case, a series of inspection work, inspection results, repair work, and the contents described in the maintenance document for each part of the equipment, as a recovery unit. The fault recovery support system according to claim 1 or 2.

4. The aforementioned work part correspondence extraction unit includes a recovery unit part estimation unit that estimates and assigns part numbers expected to be used or used in repair work by comparing the recovery unit extracted from a maintenance document containing a record of work instruction with the recovery unit extracted from a maintenance document containing a record of work performed in the same failure case, or by comparing the recovery units extracted from maintenance documents in different failure cases. The fault recovery support system according to feature 3.

5. The aforementioned work part correspondence extraction unit includes a recovery unit aggregation unit that, for all extracted recovery units, assigns work part correspondence relationships between work and parts included in the same recovery unit, using work and parts as keys, and further treats work of the same type and parts of the same type as the same, and assigns work part correspondence relationships between work and parts. The fault recovery support system according to feature 3.

6. The system includes a recommendation basis display unit that calculates the reliability of the aforementioned maintenance document based on attribute information including the document creator's years of experience, position, same-day repair completion rate, type of maintenance document, and the number of business trips required to restore the said malfunction, and that displays the maintenance document and its reliability to the user as the basis information for recommending the said work or parts. The fault recovery support system according to feature 1.

7. The recommendation matrix calculation unit calculates a recommendation score by weighting it according to the level of confidence, based on the system's operating time and the amount of data. The fault recovery support system according to feature 6.

8. The system further includes a maintenance document draft display unit that shows a draft of a maintenance document created by the user by expanding the text stored in the calculated recovery unit based on the results of the aforementioned work or parts recommendation, or the content of the recommendation accepted by the user. The fault recovery support system according to feature 3.

9. A fault recovery support method that recommends the necessary work procedures and part numbers when a malfunction occurs in equipment, The work content extraction unit extracts keywords for work content from a maintenance document, which is a history containing one or more natural documents that record the work instructed or performed to restore the equipment in the past, based on a work word table, and assigns similar work identifiers for work similar to the work in question based on the feature quantities of the keywords. The work part correspondence extraction unit extracts work part correspondences by estimating which work content a part number from a history that records the part number expected to be used when the work was instructed or the part number used when the work was performed corresponds to, and When the recommendation matrix calculation unit receives information identifying a malfunction in the equipment, it performs the steps of calculating a similar work identifier for the necessary work and a recommendation score for the part, The recovery unit recommendation unit recommends parts corresponding to similar work identifiers calculated by the recommendation matrix calculation unit, and work corresponding to parts calculated by the recommendation matrix calculation unit, using the work part correspondence relationship. A fault recovery support method characterized by comprising:

10. A fault recovery support method that recommends the necessary work procedures and part numbers when a malfunction occurs in equipment, The work content extraction unit extracts the necessary work content from maintenance documents, which are a history containing one or more documents recording the work instructed or performed to restore the equipment in the past. The work part correspondence extraction unit extracts work part correspondences by estimating which work content a part number from a history that records the part number expected to be used when the work was instructed or the part number used when the work was performed corresponds to, and When the recommendation matrix calculation unit receives information identifying a malfunction in the equipment, it performs the following steps: calculates a recommendation score for the necessary work or parts; The recovery unit recommendation unit recommends the parts corresponding to the work calculated by the recommendation matrix calculation unit, and the work corresponding to the parts calculated by the recommendation matrix calculation unit, by supplementing them using the work part correspondence relationship. The recommendation basis display unit calculates the reliability of the maintenance document based on attribute information including the document creator's years of experience, position, same-day repair completion rate, type of maintenance document, and the number of business trips required to restore the malfunction, and presents the maintenance document and its reliability to the user as the basis information for the recommendation of the work or parts. A fault recovery support method characterized by comprising:

Images