System and method for proposing operation knowledge

Abstract

To provide a system for proposing operation knowledge allowed to provide a user with appropriate maintenance knowledge.SOLUTION: A system for proposing operation knowledge 101 comprises an operating-condition searching section 42 that, if a user 80 inputs any occurrence-of-events information, proposes an operation and operation element required, a proposal-displaying section 71 that displays confidence-degree information corresponding to an operation and operation element proposed and accepts a change from the user 80, and an operating-condition presuming section 44 that, when received a change, searches a candidate for an operating condition of an operation and operation element before or after the change, based on a combination of occurrence-of-events information inputted and selects a candidate corresponding to a history accumulated in the operation-accumulating section 20.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to a task knowledge suggestion system and a task knowledge suggestion method. [Background technology]

[0002] As background art of this technical field, the following Patent Documents 1 and 2 describe techniques for presenting maintenance work procedures to workers in facilities such as plants. The descriptions in these documents are incorporated as part of the present specification. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2010-122847 A [Patent Document 2] Patent Publication No. 2021-196858 Summary of the Invention [Problem to be solved by the invention]

[0004] However, in the above-mentioned technology, there is a demand for providing users with more appropriate maintenance knowledge. The present invention has been made in consideration of the above-mentioned circumstances, and has an object to provide a work knowledge suggestion system and a work knowledge suggestion method that can provide appropriate maintenance knowledge to a user. [Means for solving the problem]

[0005] In order to solve the above problem, the work knowledge proposal system of the present invention is a work knowledge proposal system that, when it becomes necessary to perform work on equipment to be maintained, proposes a necessary work and work elements including parts or procedures related to the work, and is characterized by having: a function of inputting and outputting data to a work knowledge storage unit that stores the work, the work elements, and the work conditions that are the conditions under which the work and the work elements are necessary; a work condition search unit that, when a user inputs any occurring event information, searches for the work conditions that correspond to the inputted occurring event information and the work and work elements that are necessary corresponding to the work conditions; a proposal display unit that displays the searched work conditions, the work and the work elements, and reliability information corresponding to the work conditions, and accepts changes from the user including the addition, modification, or deletion of the work or work elements; a work condition estimation unit that, when the change is accepted, searches for and displays candidates for work conditions related to the work and work elements before or after the change based on a combination of the inputted occurring event information; and a change confirmation unit that accepts an answer from the user as to whether the candidate is valid or not, and adds the work condition for which the answer is affirmative to the work knowledge storage unit. Effect of the Invention

[0006] According to the present invention, appropriate maintenance knowledge can be provided to the user. [Brief description of the drawings]

[0007] [Figure 1] FIG. 1 is a block diagram of an operation knowledge suggestion system according to a first embodiment. [Diagram 2] FIG. 13 is a diagram illustrating an example of a data configuration of a work content table. [Diagram 3] FIG. 4 is a diagram illustrating an example of a data configuration of a work condition table. [Figure 4] FIG. 13 is a diagram illustrating an example of a data configuration of a reliability table. [Diagram 5] FIG. 13 is a diagram illustrating an example of a data configuration of a work planner table. [Figure 6] FIG. 13 is a diagram illustrating an example of a data configuration of a working word table. [Figure 7]FIG. 11 is a diagram illustrating an example of a data configuration of a working document table. [Figure 8] FIG. 13 is a diagram illustrating an example of a data configuration of a work history table. [Figure 9] FIG. 13 is a diagram illustrating an example of a matrix. [Figure 10] FIG. 13 is a diagram illustrating an example of a target task table. [Figure 11] FIG. 13 is a diagram showing an example of a proposal display screen. [Figure 12] FIG. 13 is a diagram showing an example of a counterfactual scenario confirmation screen. [Figure 13] FIG. 13 is a diagram showing an example of a change confirmation screen. [Figure 14] 13 is a flowchart of an operation content extraction process routine. [Figure 15] 13 is a flowchart of a work condition extraction processing routine. [Figure 16] 13 is a flowchart of a reliability calculation routine. [Figure 17] 13 is a flowchart of a work condition search routine. [Figure 18] 13 is a flowchart of a working condition test routine. [Figure 19] 4 is a flowchart of a working condition estimation routine. [Figure 20] FIG. 11 is a diagram showing a processing flow of a proposal display unit. [Figure 21] 13 is another diagram showing the processing flow of the proposal display unit. FIG. [Figure 22] FIG. 1 is a block diagram of a computer. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0008] [Outline of the embodiment] According to the technology applying the above-mentioned Patent Documents 1 and 2 (hereinafter referred to as the applied technology), it is believed that it is possible to realize a work procedure management system that accumulates past work contents as work records, organizes efficient work procedures from them, and presents them to workers, thereby supporting the efficiency of the workers' work. In other words, it is believed that it is possible to use the priority know-how obtained by analyzing the work logs to suggest in what order the work should progress after a certain stage. This is based on knowledge that, in a sense, records the work to be done next after the work has been completed in a corresponding manner.

[0009] The work log in the above-mentioned applied technology is a history entered by the user of the work content, work results, reasons for the work, etc., performed when responding to a malfunction. For the user to enter this information every time a malfunction is responded to, it would be a heavy workload. In addition, the user needs to have enough knowledge to be able to concisely explain the reasons for the work.

[0010] In addition, when a new task is extracted from the task log, i.e., when a task that is not as proposed is detected, knowledge of performing the new task in that situation is newly acquired. However, strictly speaking, in the above-mentioned situation, there are many possible options, such as adding a new task, modifying the proposed task, or deleting the proposed task. By appropriately repeating these choices, it becomes possible for the extracted explicit knowledge to reproduce the user's tacit knowledge.

[0011] If knowledge continues to be added without making the above-mentioned selection, incorrect knowledge or uncommon knowledge that is not incorrect will be proposed, and the user will be required to put in effort and ability to examine the proposal. In the above-mentioned applied technology, the method of assigning the reason for the work and priority to the proposed work requires the user to put in effort and ability to grasp and understand the content, and does not fundamentally solve this problem. In addition, in the above-mentioned applied technology and in the field of maintenance assumed in the embodiment described below, there is a problem that knowledge is enriched only for frequent events due to bias in the occurrence of failure events, and knowledge of rare events and unknown failures is not accumulated. Therefore, in order to eliminate such bias, it is considered preferable to provide a mechanism for actively drawing out the user's knowledge.

[0012] The embodiment described below has been devised in consideration of the above circumstances, and resolves the problems of updating and reliability of knowledge in the above-mentioned applied technology, reduces the work burden on the user and the knowledge required, and provides support for selecting the optimal operation for the event that has occurred. That is, the embodiment described below introduces a means for actively providing operation knowledge and a method for facilitating updating the knowledge and understanding the basis thereof, thereby providing the user with more appropriate maintenance knowledge.

[0013] To achieve this objective, in this embodiment, instead of managing work knowledge in a serial manner as in the above-mentioned applied technology, a parallel method is used in which the circumstances under which each work is required are recorded independently for each work. Knowledge in this embodiment is composed of work and work elements such as work procedures and parts used, which are contents that are selected flexibly in association with the work, and work conditions that define the events when each work and work element is required and the work and work elements of the previous stage. This parallel method allows knowledge to be updated flexibly and can clearly express the grounds for the execution of each work. On the other hand, the expression of knowledge becomes complicated, making it difficult to initialize or edit the knowledge directly by hand.

[0014] From a first perspective, the embodiment described below can be considered to be an apparatus including a work knowledge memory unit, a work history accumulation unit, a work planner terminal, a proposal display unit, a work condition search unit, and a change confirmation unit. Here, the work knowledge storage unit accumulates information such as work, work elements, work conditions, and reliability of the work conditions. The work history storage unit accumulates a history of executed work and work elements. The work planner terminal is a terminal operated by the user, who is the work planner. The proposal display unit displays a proposal of a work candidate and the work conditions and reliability used in the proposal when work is required by the user. The work condition search unit evaluates whether any of the work conditions stored in the work knowledge storage unit corresponds to the input content when the user inputs an event that has occurred in a situation where work is required, and proposes corresponding work and work elements.

[0015] [First embodiment] Overall Configuration of the First Embodiment FIG. 1 is a block diagram of an operation knowledge suggestion system 101 according to the first embodiment. In FIG. 1, the work knowledge suggestion system 101 includes a work knowledge storage unit 10, a work history accumulation unit 20, a work knowledge initialization unit 30, a work knowledge suggestion unit 40, and a planner terminal 70.

[0016] Here, the work knowledge storage unit 10 includes a work content table 11, a work condition table 12, a reliability table 13, a work planner table 14, and a work vocabulary table 15. The work history accumulation unit 20 includes a work document table 21 and a work history table 22. The work knowledge initialization unit 30 includes a work content extraction unit 31 and a work condition extraction unit 32. The work knowledge proposal unit 40 includes a reliability calculation unit 41, a work condition search unit 42 (work condition search process), a work condition testing unit 43, and a work condition estimation unit 44 (work condition estimation process). The planner terminal 70 includes a proposal display unit 71 (proposal display process), a counterfactual simulation confirmation unit 72, and a change confirmation unit 73.

[0017] The work knowledge suggestion system 101 suggests knowledge of maintenance work and the like for a maintenance target facility 300. The maintenance target facility 300 includes maintenance target devices 302, 304, and 306. The maintenance target facility 300 is, for example, a wind power generation system, and the maintenance target devices 302, 304, and 306 are, for example, wind power generators. In addition, the illustrated users 80-1, 80-2, ... are multiple work planners. When a situation arises where work such as repair is necessary for the maintenance target facility 300, the work planners consider and instruct the work elements such as the work to be performed and the procedure and necessary parts based on the event that has occurred. When there is no need to distinguish between the users 80-1, 80-2, ..., they are collectively referred to as "users 80."

[0018] To consider the work content, the user 80 operates the planner terminal 70 and first views the proposal display unit 71. Then, when an event that requires work occurs, the user 80 inputs the event into the planner terminal 70. The proposal display unit 71 requests a proposal for the work content from the work knowledge proposal unit 40 via the network 60, and displays the returned proposal to the user 80.

[0019] The user 80 operates the proposal display unit 71 and makes changes if he or she has any objections to the proposed work content. Depending on the change operation or input event of the user 80, the proposal display unit 71 may request the user 80 to confirm his or her knowledge using the change confirmation unit 73 or the counterfactual assumption confirmation unit 72.

[0020] In the following embodiment, the proposal of knowledge regarding maintenance work is assumed, the work being inspection and repair, the work elements being the inspection or repair procedure and the parts used for repair, and the events being the error code of the occurring failure and the model of the target machine.

[0021] Data Structure (Example of input text) Various data configurations in this embodiment will be described below. In this embodiment, the work document table 21 of the work history storage unit 20 stores work content record sentences in natural language. As examples of these work content record sentences, work content record sentences #1 and #2 are shown below. The contents of work content record sentences #1 and #2 are enclosed in quotation marks (" ").

[0022] (Work Record Document #1) "1. An E001 error occurred on 10 / 15 in the morning and the system stopped. The following work was carried out. 2. Check the operation of the noise sensor 2.1. Tap the area around the noise sensor to check for voltage changes. Result: Abnormal 2.2.Replacement of noise sensor Replacement part: NS102 3. Check the operation of board A 3.1. Check the connections (1 to 3) with the noise sensor on board A. Result: OK.

[0023] (Work Record Document #2) "1. E001 and E002 occurred late at night on December 21st. Recovery work was carried out on-site the following day. 2. Noise sensor voltage measurement result: NG -Replace the noise sensor with a new one Use NS104 3. Temperature sensor voltage measurement Result: OK 4. Check the connection of board A Check the contact of x1 to x5 (noise sensor side and temperature sensor side) of board A. Result: OK.

[0024] FIG. 2 is a diagram showing an example of the data structure of the work content table 11. As shown in FIG. The work content table 11 has a plurality of records 11R1, 11R2, ..., and each record corresponds to a keyword extracted from the work content record documents #1, #2. Each record 11R1, 11R2, ... of the work content table 11 includes case ID information 110, node number 111, keyword information 112, original document information 113, upper node ID information 114, node type information 115, and node ID information 116.

[0025] The case ID information 110 is information indicating a case of work. The above-mentioned work content record documents #1 and #2 correspond to the case of work, respectively, and "1" and "2" are assigned to both as the case ID information 110. The work content record documents #1 and #2 can be divided into multiple character strings (called reference character strings), and keywords can be extracted from each reference character string. The original document information 113 is the reference character string itself from which the keyword was extracted, and the contents of the main part of the reference character string. The node number 111 is the appearance order of the reference character string in the case ID information 110. The keyword information 112 is the contents of one or more keywords extracted from the reference character string. However, for the model information corresponding to the work content record documents #1 and #2 (for example, "E70" in record 11R8), it is included in the keyword information 112 even if there is no original document information 113. In FIG. 2, a semicolon (;) indicates a separator when there are multiple keywords.

[0026] The node ID information 116 indicates the node ID corresponding to the record. Each record in the work content table 11 may contain multiple records that have the same content or that can be safely considered to have the same content. For example, records 11R1 and 11R9 have the same content because both keyword information 112 and original text information 113 are "E001." In such a case, the node IDs of both records will be the same ("1" in the illustrated example). The method of assigning the node ID information 116 will be described in detail later.

[0027] The node type information 115 specifies a node type such as element K2 (work element), event K4, inspection K12, repair K14, etc. as the type of the record based on the keyword information 112 of the record. Here, inspection K12 and repair K14 are collectively referred to as work K10. In this embodiment, the element K2 in the node type information 115 is assumed to be a work procedure and parts used, but may also include the result of the work, priority, etc. In addition, the event K4 in the node type information 115 (see FIG. 2) in this embodiment is assumed to be an "error code" and a "model", but may also be the location conditions of the maintenance target devices 302, 304, 306 or the name of a work event such as a 10-year inspection, or may include the work content recently performed on the maintenance target devices 302, 304, 306.

[0028] The higher-level node ID information 114 indicates the node ID of another node when the node corresponding to the record is a description of another node, or when a subordinate relationship can be read on the document. For example, if the node indicates the procedure for a certain inspection work, the inspection work becomes the higher-level node. For example, the original text information 113 of record 11R3 is "tap around the noise sensor and check for voltage changes," which is considered to indicate the procedure for "checking the operation of the noise sensor" in record 11R2. Therefore, the higher-level node ID information 114 of record 11R3 is set to "2," which is the node ID information 116 of record 11R2.

[0029] Here, it is assumed that a "certain judgment content" occurs when (one or more) "certain condition" is satisfied. A "certain judgment content" is, for example, a judgment content such as "a certain task must be performed." The combination of (one or more) "certain condition" and "certain judgment content" is called an "operation condition." In such a case, the higher-level node of the "certain judgment content" becomes the "certain condition." The method of assigning higher-level node ID information 114 will be described in detail later.

[0030] FIG. 3 is a diagram showing an example of the data configuration of the work condition table 12. As shown in FIG. 3, the work condition table 12 includes a plurality of records (work conditions) 12R1, 12R2, ..., with one record corresponding to one "work condition." Each record 12R1, 12R2, ... includes node ID information 120, from ID information 121 (work), to ID information 122 (work element), work field information 123, and reliability information 124.

[0031] The toID information 122 in each record is a node ID in the work content table 11 (see FIG. 2), and the fromID information 121 is its higher-level node ID. Each record in the work condition table 12 can form a knowledge graph when the nodes corresponding to the fromID information 121 and the toID information 122 are connected by edges. The node ID information 120 is unique information that identifies the work condition related to the record.

[0032] Working field information 123 indicates the technical field of the working conditions related to the record, i.e., a field code such as "electrical", "mechanical", "temperature", etc. In other words, working field information 123 indicates the technical field based on keyword information 112 (see FIG. 2) for both the nodes of fromID information 121 and toID information 122. The technical field is information corresponding to classification information 153 of word information 150 (see FIG. 6) described later. Reliability information 124 indicates the reliability of the working conditions related to the record. In other words, the larger the value of reliability information 124, the greater the reliability of the working conditions related to the record.

[0033] FIG. 4 is a diagram showing an example of the data configuration of the reliability table 13. As shown in FIG. 4, the reliability table 13 has a plurality of records (reliability history information) 13R1, 13R2, ..., and one record represents the result of an "evaluation" by a user 80 of each work condition (record) in the work condition table 12 (see FIG. 3). Each record in the reliability table 13 includes node ID information 130 (work content), input date information 131, planner ID information 132, status information 133, and judgment information 134.

[0034] The node ID information 130 is equal to the node ID information 120 (see FIG. 3) of the working condition evaluated by the user 80. The input date information 131 is the date on which the user 80 input the record in question in the reliability table 13. The planner ID information 132 is the identification information of the user who input the record in question. The status information 133 is either "Proposal" or "Counterfactual Check". Here, "Proposal" indicates that the working condition in question is "proposed". Furthermore, "Counterfactual Check" indicates that the working condition in question is one for which "counterfactual check" is required.

[0035] The judgment information 134 indicates the judgment result of the work conditions related to the record, and can be one of "affirmative," "negative," or "pending." Here, "affirmative" means that "the work conditions are correct," "negative" means that "the work conditions are incorrect," and "pending" indicates that the judgment result has not yet been determined.

[0036] FIG. 5 is a diagram showing an example of a data configuration of the work planner table 14. As shown in FIG. 5, the work planner table 14 has a plurality of records 14R1, 14R2, ..., and one record corresponds to one user 80. Each record in the work planner table 14 includes planner ID information 140, job title information 141, and specialty information 142. The planner ID information 140 is information that uniquely identifies the user 80 related to the record. The job title information 141 indicates the job title of the user 80. The specialty information 142 stores the field code information in which the user 80 is skilled. The field code information is the same as the technical field in the work field information 123 of the work condition table 12 (see FIG. 3).

[0037] FIG. 6 is a diagram showing an example of the data structure of the working word table 15. As shown in FIG. As previously exemplified as work content record sentences #1 and #2, in this embodiment, the user describes, in a sentence in natural language, the contents of the maintenance and inspection work that has occurred for the maintenance target equipment 300. The work word table 15 is a table that defines important terms among the words that may appear in the sentence in natural language.

[0038] 6, the working word table 15 includes a plurality of records 15R1, 15R2, ..., each of which corresponds to a word. Each record includes word information 150, label information 151, node type information 152, and classification information 153.

[0039] The label information 151 indicates the type of the word information 150, and includes, for example, the following: ·machine: Indicates the model number of the equipment 302, 304, or 306 to be maintained. · target: Indicates the part of the equipment where maintenance work or inspection was performed. ·inspect: indicates the content of the inspection carried out on that part · repair: indicates the repair or replacement for that part · error: Indicates an event such as an error code that occurred. · parts: Indicates the part model number, etc.

[0040] Here, in the label information 151, "machine" and "target" are called "object-based labels", "inspect" and "repair" are called "predicate-based labels", and "error" and "parts" are called "single-based labels". In addition, among the words shown in the word information 150, those with object-based labels in the label information 151 are called "object-based words", those with predicate labels are called "predicate words", and those with single-based labels are called "single-based words".

[0041] The node type information 152 indicates the type of the word information 150 in the record, and the types include "event," "inspection," "repair," and "element." Here, "event" means that the word information 150 "represents an event that has occurred in the maintenance target facility 300." Furthermore, "inspection" means that the word information 150 "represents an inspection to be performed in the maintenance target facility 300." Furthermore, "repair" means that the word information 150 "represents a repair to be performed in the maintenance target facility 300." Furthermore, "element" means that the word information 150 "represents an element related to an event, inspection, or repair."

[0042] The classification information 153 indicates the field code when a specialized field related to the content of the word information 150 exists. The field code information is the same as the technical field in the work field information 123 of the work condition table 12 (see FIG. 3).

[0043] FIG. 7 is a diagram showing an example of the data configuration of the working document table 21. As shown in FIG. 7, the working document table 21 includes a plurality of records 21R1, 21R2, ..., and each record corresponds to a "maintenance and inspection work" that has occurred for the maintenance target equipment 300 (see FIG. 1). Each record includes case ID information 210, occurrence date information 211, planner ID information 212, target model information 213, and work content information 214.

[0044] The case ID information 210 is unique information for identifying the maintenance and inspection work related to the record. The occurrence date information 211 indicates the date and time when the maintenance and inspection work occurred. The planner ID information 212 is information for identifying the user 80 who is the planner of the maintenance and inspection work, and corresponds to the planner ID information 140 in the work planner table 14 (see FIG. 5). The target model information 213 is information indicating any one of the maintenance target devices 302, 304, 306 on which the maintenance and inspection work occurred. The work content information 214 is a sentence that describes the content of the above-mentioned maintenance and inspection work in a character string of natural language. In the illustrated example, the records with the case ID information 210 of "1" and "2" respectively store the above-mentioned work content record sentences #1 and #2 as the work content information 214.

[0045] FIG. 8 is a diagram showing an example of the data configuration of the work history table 22. As shown in FIG. The work history table 22 includes a plurality of records (history) 22R1, 22R2, ..., and each record corresponds to the above-mentioned maintenance and inspection work, similar to the working document table 21 (see FIG. 7). Each record includes case ID information 220, planner ID information 221, occurrence date information 222, occurring event information 223, planned work information 224, and effective work information 225.

[0046] The case ID information 220, planner ID information 221, and occurrence date information 222 are respectively similar to the case ID information 210, planner ID information 212, and occurrence date information 211 in the above-mentioned case ID information 210 (see FIG. 7). The occurred event information 223 indicates one or more events that occurred in response to the maintenance and inspection work related to the record. More specifically, the occurred event information 223 includes one or more pieces of node ID information 116 of a record whose node type information 115 in the work content table 11 (see FIG. 2) is "event."

[0047] The planned work information 224 indicates inspection work that can be performed on a location where an abnormality has been found, repair work that can be performed to recover from the abnormality, etc. More specifically, one or more pieces of node ID information 116 of records in which the node type information 115 in the work content table 11 (see FIG. 2) is "inspection" or "repair" are included in the planned work information 224. The valid work information 225 is the node ID information 116 of the work that has actually been performed out of the works listed in the planned work information 224.

[0048] FIG. 9 is a diagram showing an example of the matrix MTC created by the work condition extraction unit 32. As shown in FIG. Each row of the matrix MTC corresponds to the toID information 122 of the work condition table 12 (see FIG. 3). Also, each column of the matrix MTC corresponds to the fromID information 121. The numerical value of each element of the matrix MTC represents a numerical value such as the number of occurrences corresponding to the fromID information 121 which is the column and the toID information 122 which is the row.

[0049] FIG. 10 is a diagram showing an example of the target task table 24. As shown in FIG. The noted task table 24 comprises a plurality of records 24R1, 24R2, ..., and each record includes occurring event information 240, case ID information 241, and task presence / absence information 242. The occurring event information 240 indicates one or more node IDs that have occurred. The case ID information 241 is a case ID that corresponds to this occurring event information 240. The task presence / absence information 242 indicates whether or not a task has been performed for the occurring event information 240.

[0050] <Work knowledge initialization section 30> The work content extraction unit 31 shown in Fig. 1 acquires the work document table 21 (see Fig. 7) in the work history accumulation unit 20 and the work vocabulary table 15 (see Fig. 6) in the work knowledge storage unit 10. Next, the work content extraction unit 31 extracts information on the maintenance and inspection work included in the work content information 214 (for example, the above-mentioned work content record documents #1 and #2), and stores the contents in the work content table 11 (see Fig. 2). In addition, the work condition extraction unit 32 generates the above-mentioned matrix MTC (see Fig. 9) and the like.

[0051] <Planner terminal 70> FIG. 11 is a diagram showing an example of a proposal display screen V71 displayed on the proposal display section 71. As shown in FIG. The proposal display unit 71 acquires a proposal or confirmation of work content from the work knowledge proposal unit 40 via the network 60, and displays the content to the user 80 on the proposal display screen V71. In Fig. 11, the proposal display screen V71 includes an event input unit 710, a work content display unit 711, a correction input button 712, an add work button 713, a delete work button 714, a proposal reason display unit 715, and a plan completion button 716.

[0052] In the event input section 710, the user 80 inputs the details of an event that has occurred in the maintenance target facility 300. In the work content display section 711, a list of work contents proposed to the user 80 in response to the event that has occurred is displayed. In the illustrated example, "temperature sensor inspection" and "board inspection" are included in the list of work contents. The user 80 can modify the contents of the list by pressing a modification input button 712.

[0053] The add work button 713 commands the work content display section 711 to add a new work. The delete work button 714 commands the deletion of any of the works displayed in the work content display section 711. The proposal basis display section 715 shows the proposal basis, etc. of any of the work contents listed in the work content display section 711. In the illustrated example, the proposal basis display section 715 shows the proposal basis, etc. of "infrastructure inspection".

[0054] FIG. 12 is a diagram showing an example of a counterfactual scenario confirmation screen V72 displayed in the counterfactual scenario confirmation section 72. As shown in FIG. 12, the counterfactual scenario confirmation screen V72 includes a work content display section 720, a proposal basis display section 721, and a hold button 724. Furthermore, the proposal basis display section 721 includes an affirmative button 722 and a negative button 723.

[0055] In the example of FIG. 12, the counterfactual scenario confirmation screen V72 proposes a work content of "Infrastructure inspection." The work content display unit 720 displays the hypothetical proposed content, which is the hypothetical proposed work content. The proposal basis display unit 721 extracts and displays the work conditions, etc. used in the hypothetical proposed content from the work condition table 12 (see FIG. 3). The affirmative button 722, the negative button 723, and the hold button 724 are used to input command information of "affirmative," "negative," and "hold," respectively. When any of these buttons is pressed, the processing of the counterfactual scenario confirmation unit 72 ends, and the display screen transitions to the screen of the proposal display unit 71.

[0056] FIG. 13 is a diagram showing an example of a change confirmation screen V73 displayed by the change confirmation unit 73. As shown in FIG. 13, the change confirmation screen V73 includes a change content display section 730, a proposal basis amendment display section 731, and a hold button 734. Furthermore, the proposal basis amendment display section 731 includes an affirmative button 732 and a negative button 733.

[0057] The change content display unit 730 displays the changes made by the user 80 via the proposal display unit 71. The proposal basis amendment proposal display unit 731 displays the work conditions estimated by the work conditions estimation unit 44. The affirmative button 732, negative button 733, and reserve button 734 are used to input command information of "affirmative", "negative", and "reserved", respectively. When any of these buttons is pressed, the processing of the change confirmation unit 73 ends, and the display screen transitions to the screen of the proposal display unit 71.

[0058] FIG. 14 is a flowchart of a task content extraction processing routine executed in task content extraction unit 31. (S3102: Data acquisition) 14, when the process proceeds to step S3102, a data acquisition process is executed. That is, the work content extraction unit 31 reads the working document table 21 (see FIG. 7) and the working word table 15 (see FIG. 6) and acquires their contents.

[0059] (S3104: String split) Next, when the process proceeds to step S3104, the operation content extraction unit 31 executes character string division. The contents of this character string division are as follows. (1) One unprocessed record is selected as a processing target from the working document table 21. This record is called a "processing target record 21Ra" (not shown). (2) The work content information 214 in the record 21Ra to be processed (for example, the above-mentioned work content record sentences #1 and #2) is divided into multiple character strings using "punctuation marks," "line breaks," and "delimiters" as boundaries. A "delimiter" refers to a character string that separates a sentence, such as "1." or "·." Each divided character string is referred to here as a "substring." (3) For each partial character string, word information 150 included in the working word table 15 (see FIG. 6) and label information 151 corresponding to the word information 150 are obtained.

[0060] (4) Each substring is divided into shorter strings according to the label information 151 of the extracted word information 150. The divided strings are called "reference strings." Here, a reference string is either a "string containing a pair of object-like words and predicate-like words" (hereinafter referred to as an "object-predicate reference string") or a "string containing a single standalone word" (hereinafter referred to as an "standalone reference string").

[0061] As described above, the "object labels" are "machine" and "target," the "predicate labels" are "inspect" and "repair," and the "single labels" are "error" and "parts." Furthermore, the "object words," "predicate words," and "single words" are words that correspond to these labels.

[0062] (S3106: Update work content table) Next, when the process proceeds to step S3106, the work content extraction unit 31 updates the work content table 11 based on the contents of the reference character string. The contents of this update process are as follows. (1) Creation of records, etc. The work content extraction unit 31 creates a record in the work content table 11 (see FIG. 2) corresponding to each reference character string. Then, the work content extraction unit 31 writes the case ID information 110 according to the work content information 214 (for example, the above-mentioned work content record sentences #1 and #2) in which the reference character string was included, and writes the node number 111 according to the appearance order of the reference character strings.

[0063] (2) Setting of original text information 113, etc. The work content extraction unit 31 judges whether or not each reference character string contains a "child element symbol." Here, a "child element symbol" is a character string such as a symbol that is part of the above-mentioned "delimiter symbol," such as "·" or "1.1.", and indicates that it is a child element item.

[0064] (2A) If a "child element symbol" is included in the reference character string, the work content extraction unit 31 deletes this "child element symbol" and writes it as original text information 113. Furthermore, in this case, the node type information 115 is set to "element".

[0065] For example, the above-mentioned work content record text #1 contains the character string "2.1. Tap the area around the noise sensor and confirm the voltage change", which is an example of a reference string. Since "2.1." is a child element symbol in this reference string, it is deleted and the remaining character string "Tap the area around the noise sensor and confirm the voltage change" is written as the original text information 113 of record 11R3 (see FIG. 2). And the node type information 115 of record 11R3 is set to "element".

[0066] (2B) On the other hand, if the "child element symbol" is not included in the reference character string, the work content extraction unit 31 writes the contents of the reference character string as original text information 113. In this case, the node type information 115 is currently undetermined.

[0067] (3) Setting keyword information 112 When the original text information 113 is an independent reference character string (a character string including a single independent word), the work content extraction unit 31 writes the word into the keyword information 112 of the record in the work content table 11. For example, in record 11R1 in Fig. 2, the original text information 113 is "E001", which is an independent reference character string, so the work content extraction unit 31 writes the word "E001" as is into the keyword information 112 of record 11R1.

[0068] On the other hand, when the original text information 113 is an object-predicate reference character string, the object words and predicate words contained therein are written into the keyword information 112 of the corresponding record in the task content table 11. For example, in record 11R2 in Fig. 2, the original text information 113 is "Confirming the operation of the noise sensor," of which "noise sensor" is an object word and "confirm" is a predicate word. Therefore, "noise sensor" and "confirm" are written as the keyword information 112.

[0069] (4) Setting node type information 115 As described above, when a reference string includes a "child element symbol," the node type information 115 in the corresponding record is set to "element." For other records, the work content extraction unit 31 determines the node type information 115 based on whether the reference string is an independent reference string or an object-predicate reference string.

[0070] (4A) In the case of a standalone reference string When the original text information 113 is an independent reference character string, it contains an independent word. The work content extraction unit 31 extracts node type information 152 of a record corresponding to the independent word in the work word table 15 (see FIG. 6). Then, the content of the extracted node type information 152 is written to node type information 115 of a record corresponding to the original text information 113 in the work content table 11 (see FIG. 2).

[0071] For example, in record 11R1 in Fig. 2, original text information 113 is "E001," which is an independent word itself. According to record 15R1 in working word table 15 (see Fig. 6), node type information 152 of "E001" is "event." Therefore, "event" is written in node type information 115 in record 11R1 in Fig. 2.

[0072] (4B) In the case of object-predicate reference strings When the original text information 113 is an object-predicate reference character string, it includes object words and predicate words. As described above, the label information 151 (see FIG. 6) corresponding to the predicate words is either "inspect" or "repair". When the label information 151 corresponding to the predicate words is "inspect", "inspection" is written in the node type information 115 of the work content table 11 (see FIG. 2). On the other hand, when the label information 151 is "repair", "repair" is written in the node type information 115 in the work content table 11.

[0073] For example, in record 11R2 in Fig. 2, the original text information 113 is "Check the operation of the noise sensor", which includes the predicate word "check". Meanwhile, in record 15R1 in Fig. 6, the label information 151 of "check" is "inspect". Therefore, in record 11R2 in Fig. 2, "inspection" is written in the node type information 115.

[0074] 2, the original text information 113 is "Replacement of noise sensor", of which "replacement" is a predicate word. According to record 15R10 in Fig. 6, the label information 151 of "replacement" is "repair". Therefore, "repair" is written in the node type information 115 in record 11R4 in Fig. 2.

[0075] (5) Obtaining the upper node number (5.1) Upper node number for "repair" Letting p and q be any natural numbers, the work content extraction unit 31 determines the upper node number (not shown) of the record 11Rp (not shown) whose node type information 115 is "repair" as follows. One or more records 11Rq (not shown) that have the same case ID information 110 as the record 11Rp, have a node number 111 smaller than the node number 111 of the record 11Rp, and have node type information 115 that is "inspection" are searched for. Among the retrieved record or records 11Rq, the node number 111 of the record with the largest node number 111 is the higher node number of the record 11Rp.

[0076] 2, the node type information 115 of record 11R4 is "repair", and record 11Rq that satisfies the above-mentioned condition becomes record 11R2. Therefore, the upper node number of record 11R4 is set to "2", which is the node number 111 of record 11R2.

[0077] (5.2) Upper node number for "element" Furthermore, the work content extraction unit 31 determines the upper node number of the record 11Rp (not shown) whose node type information 115 is "element" as follows. · One or more records 11Rq (not shown) are searched for that have the same case ID information 110 as record 11Rp, have a node number 111 smaller than the node number 111 of record 11Rp, and have node type information 115 that is "inspection" or "repair". Among the retrieved record or records 11Rq, the node number 111 of the record with the largest node number 111 is the higher node number of the record 11Rp.

[0078] (6) Setting the node type information 115 corresponding to the model information The work content extraction unit 31 also extracts one or more records from the work content table 11 that have case ID information 110 equal to the case ID information 210 of the processing target record 21Ra (the record being processed in the working document table 21). If the original text information 113 of the extracted record contains the target model information 213 in the corresponding records 21R1, 21R2, ..., the node type information 115 in the record is set to "event".

[0079] (S3108: All records processed?) 14, when the process proceeds to step S3108, the work content extraction unit 31 judges whether or not the process of steps S3104 and S3106 has been executed for all records 21R1, 21R2, ... in the working document table 21. If the judgement here is "No", the process returns to step S3104, and the work content extraction unit 31 repeats the process of steps S3104 and S3106 for any unprocessed records in the working document table 21. On the other hand, if the judgement here is "Yes", the process proceeds to step S3110.

[0080] (S3110: Setting node ID information, upper node ID information) (1) In step S3110, the work content extraction unit 31 assigns node ID information 116 to each record in the work content table 11 in the following procedure. The node ID information 116 of the first record 11R1 that occurs is set to "1". For the second or subsequent newly occurring records, it is determined whether the keyword information 112 of that record (the set of words listed in the keyword information 112) and the keyword information 112 of other records that occurred in the past satisfy a predetermined "identical / similar relationship." If the determination result is positive, the node ID information 116 of the newly generated record is set to a value common to other records that satisfy the "identical / similar relationship".

[0081] If the determination result is negative, new node ID information 116 is assigned to the newly generated record. The "identical / similar relationship" can be determined by calculating the similarity using a known method such as "Word2Vec". In other words, if the calculated similarity is equal to or greater than a predetermined threshold, it is determined that the "identical / similar relationship" is satisfied, and otherwise it is determined that the "identical / similar relationship" is not satisfied.

[0082] (2) In addition, in step S3110, the work content extraction unit 31 sets the upper node ID information 114 for each record in the work content table 11 that has the above-mentioned "upper node number." In other words, when the record corresponding to the "upper node number" of a certain record 11Rp (not shown) is "record 11Rq (not shown)," the work content extraction unit 31 writes the node ID information 116 of record 11Rq into the upper node ID information 114 of record 11Rp.

[0083] FIG. 15 is a flowchart of a work condition extraction processing routine executed in the work condition extraction unit 32. (S3202: Data acquisition, matrix initialization) 15, when the process proceeds to step S3102, data acquisition processing is executed. That is, the work condition extraction unit 32 acquires the work content table 11 (see FIG. 2). In addition, three matrices MTA, MTB, and MTC (not shown) that are null matrices are generated.

[0084] (S3204: ID extraction) Next, when the process proceeds to step S3204, the work condition extraction unit 32 performs an ID extraction process. The contents of this process are as follows. (1) The work condition extraction unit 32 acquires one piece of case ID information 110 that has not been subjected to the processing in step S3206, which will be described later, from among the multiple pieces of case ID information 110 included in the work content table 11 (see FIG. 2). The acquired case ID information 110 is referred to as "instruction ID information 110a" (not shown).

[0085] (2) Among the instruction ID information 110a, all node ID information 116 whose corresponding node type information 115 is “event”, “inspection”, “repair”, or “element” is acquired. The acquired node ID information 116 for “event”, “inspection”, “repair”, and “element” are referred to as event ID information 116J, inspection ID information 116T, repair ID information 116S, and element ID information 116Y, respectively (none of which are shown in the figure).

[0086] (S3206:Add matrix) Next, when the process proceeds to step S3206, the work condition extraction unit 32 performs a matrix addition process. The details of this process are as follows. (1) All inspection ID information 116T is added as "rows" of a matrix MTA, all event ID information 116J is added as "columns" of the matrix MTA, and all elements of the matrix MTA are initialized to "0".

[0087] (2) All repair ID information 116S is added as "rows" of the matrix MTB, and all "combinations of upper node ID information 114 of these repair ID information 116S and event ID information 116J" are added as "columns" of the matrix MTB. Then, all elements of the matrix MTB are initialized to "0".

[0088] (3) All element ID information 116Y is added as "rows" of the matrix MTC, and all "combinations of upper node ID information 114 of these element ID information 116Y and event ID information 116J" are added as "columns" of the matrix MTC. Then, all elements of the matrix MTC are initialized to "0".

[0089] (S3208: All cases dealt with?) Next, when the process proceeds to step S3208, the work condition extraction unit 32 determines whether or not all of the case ID information 110 in the work content table 11 (see FIG. 2) have been selected as the above-mentioned instruction ID information 110a. If the determination here is "No", the process returns to step S3204, and the work condition extraction unit 32 repeats the processes of steps S3204 and S3206 for any unprocessed case ID information 110 in the work content table 11. On the other hand, if the determination here is "Yes", the process proceeds to step S3210.

[0090] (S3210: Remove duplicate rows and columns) In step S3210, the work condition extraction unit 32 deletes duplicated "rows" and duplicated "columns" in the above-mentioned matrices MTA, MTB, and MTC.

[0091] (S3212: ID re-extraction process) Next, when the process proceeds to step S3212, the work condition extraction unit 32 performs an ID re-extraction process. The contents of this process are as follows. (1) The work condition extraction unit 32 acquires one piece of case ID information 110 that has not been subjected to the processing in step S3214 described later, from among the case ID information 110 included in the work content table 11 (see FIG. 2). The acquired case ID information 110 is referred to as “instruction ID information 110b” (not shown).

[0092] Of the node ID information 116 corresponding to the instruction ID information 110b, all node ID information 116 whose corresponding node type information 115 is "event", "inspection", "repair", or "element" is acquired. As in step S3204 described above, the node ID information 116 acquired for "event", "inspection", "repair", and "element" are referred to as event ID information 116J, inspection ID information 116T, repair ID information 116S, and element ID information 116Y, respectively (none of which are shown in the figure).

[0093] (S3214: Matrix editing process) Next, when the process proceeds to step S3214, the work condition extraction unit 32 performs a matrix editing process. The details of the process are as follows. (1) The inspection ID information 116T and the event ID information 116J corresponding to the instruction ID information 110b are obtained, and "1" is added to the elements of the above-mentioned matrix MTA that correspond to the inspection ID information 116T and the event ID information 116J.

[0094] (2) Obtain repair ID information 116S corresponding to instruction ID information 110b and "the combination of upper node ID information 114 of this repair ID information 116S and event ID information 116J." Then, add "1" to the element of the above-mentioned matrix MTB that corresponds to the repair ID information 116S and "the combination of upper node ID information 114 of the repair ID information 116S and event ID information 116J."

[0095] (3) Obtain element ID information 116Y corresponding to instruction ID information 110b and "the combination of upper node ID information 114 of this element ID information 116Y and event ID information 116J." Then, add "1" to the element of the above-mentioned matrix MTC that corresponds to the element ID information 116Y and "the combination of upper node ID information 114 of the element ID information 116Y and event ID information 116J."

[0096] (S3216: All cases dealt with?) Next, when the process proceeds to step S3216, the work condition extraction unit 32 determines whether or not all of the case ID information 110 in the work content table 11 (see FIG. 2) have been selected as the above-mentioned instruction ID information 110b. If the determination here is "No", the process returns to step S3212, and the work condition extraction unit 32 repeats the processes of steps S3212 and S3214 for any unprocessed node ID information 116 in the work content table 11. On the other hand, if the determination here is "Yes", the process proceeds to step S3218.

[0097] (S3218: Select unprocessed matrix) Next, when the process proceeds to step S3218, the work condition extraction unit 32 selects one matrix that has not yet been processed in steps S3220 and S3222 described below from among the matrices MTA, MTB, and MTC. The selected matrix is ​​called a selected matrix MTX (not shown).

[0098] (S3220: Select unprocessed rows) Next, when the process proceeds to step S3220, the work condition extraction unit 32 selects one "row" that has not yet been processed in step S3222 described later from the selection matrix MTX. The selected "row" is called a selected row MTX-L (not shown).

[0099] (S3222: Processing for selected rows) Next, when the process proceeds to step S3222, the work condition extraction unit 32 executes a selected row corresponding process, the contents of which are as follows. (1) A threshold value Th (not shown) is determined based on the elements included in the selected row MTX-L. Here, the threshold value Th can be, for example, the median value of the elements included in the selected row MTX-L. (2) Among the elements included in the selected row MTX-L, elements that exceed a threshold Th are extracted, and one or more columns of the selection matrix MTX that include the extracted elements are obtained. The obtained columns are called obtained columns MTX-C.

[0100] (3) Work condition extraction unit 32 adds a new record to work condition table 12 (see FIG. 3), and writes unique information as the node ID information 120. The added record is called added record 12Ra (not shown). (4) The work condition extraction unit 32 sets the contents of the additional record 12Ra as follows. The node ID information 116 (see FIG. 2) corresponding to the selected row MTX-L is written into the toID information 122 of the added record 12Ra. One or more pieces of node ID information 116 corresponding to the obtained column MTX-C are written to the fromID information 121 of the added record 12Ra.

[0101] (S3224: All rows processed?) Next, when the process proceeds to step S3224, the work condition extraction unit 32 determines whether or not all rows in the selection matrix MTX have been selected as selected rows MTX-L. If the determination here is "No", the process returns to step S3220, and the work condition extraction unit 32 repeats the processes of steps S3220 and S3222 for any unprocessed "rows" in the selection matrix MTX. On the other hand, if the determination here is "Yes", the process proceeds to step S3226.

[0102] (S3226: All matrices processed?) Next, when the process proceeds to step S3226, the work condition extraction unit 32 determines whether or not all of the matrices MTA, MTB, and MTC have been selected as the selected matrix MTX. If the determination here is "No", the process returns to step S3218, and the work condition extraction unit 32 repeats the processes of steps S3218 to S3222 for any unprocessed matrices. On the other hand, if the determination here is "Yes", the process of this routine ends.

[0103] Through the above processing, the work condition extraction unit 32 extracts records related to "inspection," "repair," or "element" from the work content table 11. Then, the work condition extraction unit 32 estimates, as a work condition, when an event occurs in which the work content related to the extracted record is performed, and stores this in the work condition table 12.

[0104] Generally, when there is not enough data, it is difficult to estimate the work conditions for the maintenance target facility 300 because there are a huge number of combinations of events to be considered. In contrast, the above-mentioned work condition extraction processing routine can simplify the estimation by limiting the combinations to be considered by referring to the upper node ID information 114 in the work content table 11.

[0105] The above process starts with the process of initializing a matrix that lists candidates for combinations of node ID information 116 that can become fromID information 121 in work condition table 12 for each case ID information 110 in work content table 11. When generating the matrix, the following assumptions AS1, AS2, and AS3 are introduced into the work conditions.

[0106] Assumption AS1: For a node corresponding to toID information 122 in the work condition table 12, if the node type information 115 of that node is "inspection", then the fromID information 121 is a node whose node type information 115 is "event". · Assumption AS2: For the node corresponding to the toID information 122 in the work condition table 12, if the node type information 115 of the node is “repair”, the fromID information 121 becomes “the combination of the upper node ID information 114 and the node type information 115 of which is an event”. · Assumption AS3: For a node corresponding to toID information 122 in the work condition table 12, if the node type information 115 of that node is “element”, the fromID information 121 becomes “a combination of its upper node ID information 114 and a node whose node type information 115 is an event”.

[0107] In the above process, according to this assumption, all elements of matrices MTA, MTB, and MTC, in which "candidate combinations of node ID information 116" that can become each fromID information 121 are the columns and toID information 122 are the rows, are initialized to "0". In addition, duplicates in each row or column are removed. The matrices MTA, MTB, and MTC initialized in this way are constructed so as to correspond to each assumption. As a result, the matrix MTC becomes, for example, as shown in FIG. 9.

[0108] Here, when assuming the result of "inspection" as the task element, the above-mentioned assumption AS2 may be appropriately modified, for example, as in the following assumption AS2'. · Assumption AS2': For the node corresponding to the toID information 122 in the work condition table 12, if the node type information 115 of the node is "repair", then the combination is "a node whose upper node ID information 114 and node type information 115 are an event or work element".

[0109] Next, for each of the matrices MTA, MTB, MTC, for each case ID information 110 in the work content table 11, a check is made to see whether the node ID information 116 contains a combination of the corresponding row and column, and if so, "1" is added to the element value of the matrix. An appropriate threshold value Th, such as a median, is determined for each row for the element values ​​thus tallied, i.e., the frequency of occurrence of each row and column combination, and row and column combinations that are equal to or greater than the threshold value Th are stored in the work condition table 12 as toID information 122 and fromID information 121, respectively.

[0110] 9, for example, if the threshold value Th is "2", the following work conditions are obtained: toID information 122 is "3" and fromID information 121 is "2", toID information 122 is "5" and fromID information 121 is "4", and toID information 122 is "5" and fromID information 121 is "4" and "8". However, the example shown in Fig. 9 is not particularly related to the above-mentioned work content record sentences #1 and #2.

[0111] <Work knowledge proposal department 40> Returning to Fig. 1, when the user 80 requests a work suggestion on the suggestion display unit 71, the work knowledge suggestion unit 40 acquires the work condition table 12 and searches for the necessary work content. Then, according to the search result, the work knowledge suggestion unit 40 suggests to the user 80 any of the work contents recorded in the work content table 11. Furthermore, the work knowledge suggestion unit 40 judges whether to execute a confirmation for the user 80 to actively expand his / her knowledge.

[0112] The reliability calculation unit 41 acquires the reliability table 13 (see FIG. 4) and calculates the reliability for each node ID information 120 in the work condition table 12 as the percentage of the work condition content that does not differ from the tacit knowledge of the user 80, i.e., each work planner. The reliability calculation unit 41 then stores the calculated reliability in the reliability information 124 in the work condition table 12 (see FIG. 3). In addition, the reliability calculation unit 41 acquires the work planner table 14 and the like in order to weight the judgment of each work planner in the reliability calculation process.

[0113] FIG. 16 is a flowchart of a reliability calculation routine executed in the reliability calculation unit 41. (S4102: Data acquisition) 16, when the process proceeds to step S4102, a data acquisition process is executed. That is, the reliability calculation unit 41 acquires the contents of the reliability table 13 (see FIG. 4).

[0114] (S4104: Planner response processing) Next, when the process proceeds to step S4104, the reliability calculation unit 41 executes a planner response process, the contents of which are as follows. (1) In the reliability table 13, obtain a record whose status information 133 is other than "Proposal." (2) Among the acquired records other than "Proposed", the ratio of records whose judgment information 134 is other than "Pending" is calculated for each planner ID information 132. The calculated ratio is called the "KM contribution" of each planner ID. The KM contribution may be written in the work planner table 14 (see FIG. 5).

[0115] (3) For each record in the work history table 22 (see FIG. 8), a valid work ratio, which is the ratio of the planned work information 224 included in the valid work information 225, is calculated. (4) Then, for each piece of planner ID information 221, an average value of the effective work ratio is calculated, and the result is called the “plan effective rate” in the work planner table 14 corresponding to that planner ID.

[0116] (S4106: Get record) Next, when the process proceeds to step S4106, the reliability calculation unit 41 executes a record acquisition process. That is, one record for which the process of step S4108 has not been executed is acquired from the work condition table 12 (see FIG. 3). The acquired record is hereinafter referred to as a record to be processed 12Rp (not shown).

[0117] (S4108: Reliability calculation) Next, when the process proceeds to step S4108, the reliability calculation unit 41 executes a reliability calculation process, the contents of which are as follows. (1) From among the records in the reliability table 13 (see FIG. 4), all records having the same node ID information 130 as the node ID information 120 in the processing target record 12Rp are obtained.

[0118] (2) The records of the acquired reliability table 13 are classified by planner ID information 132, and for each planner ID information 132, a record with the latest input date information 131 is identified. The node ID information 130 of the identified record is called a planner-related identification number Cq (not shown). Therefore, there are as many planner-related identification numbers Cq as there are corresponding planner IDs. In the example of FIG. 4, the planner-related identification number Cq is one of "C1" to "C12."

[0119] (3) As shown in the work condition table 12 (see FIG. 3), the planner-related identification number Cq (for example, any one of C1 to C12) is associated with fromID information 121 and toID information 122. Then, the keyword information 112 of the work content table 11 (see FIG. 2) corresponding to the planner-related identification number Cq is acquired. The acquired keyword information 112 includes any one of the word information 150 of the work word table 15. Then, the classification information 153 corresponding to the word information 150 included in the keyword information 112 is acquired. Next, the rate at which the acquired classification information 153 matches the specialty field information 142 (see FIG. 5) corresponding to the corresponding planner ID is calculated. The calculated rate is called the "field suitability" corresponding to the planner ID.

[0120] (4) The KM contribution degree, plan effectiveness rate, and field suitability degree corresponding to the planner ID are summed up using an arbitrary formula, and the result is set as the “weight” of each planner ID information 140. (5) For each planner ID, the proportion of judgment information 134 (see FIG. 4) that is “positive” is weighted by the “weight”, and the calculation result is written into the work condition table 12 as reliability information 124 of the record corresponding to the above-mentioned planner corresponding identification number Cq.

[0121] (6) For each planner ID, the proportion of judgment information 134 (see FIG. 4) that is “negative” is weighted by a “weight”. If the result of this calculation is equal to or greater than a predetermined threshold, the reliability information 124 of the record corresponding to the planner corresponding identification number Cq in the work condition table 12 (see FIG. 3) is set to “0 (abolished)”.

[0122] (S4110: All records processed?) Next, when the process proceeds to step S4110, the reliability calculation unit 41 judges whether or not the process of step S4108 has been executed for all records in the work condition table 12 (see FIG. 3). If the judgement here is "No", the process returns to step S4106, and the reliability calculation unit 41 repeats the processes of steps S4106 and S4108 for any unprocessed records. On the other hand, if the judgement here is "Yes", the process of this routine ends.

[0123] In this way, according to this routine, the "weight" of each user 80 can be found based on the KM contribution, plan effectiveness rate, and field suitability. Of these, the KM contribution and plan effectiveness rate should be calculated for each user 80 who is the work planner, and the "field suitability" should be calculated for each combination of work conditions and user 80. The KM contribution is calculated while there is variation among users 80 in their IT skills, how busy they are, their attitudes toward knowledge management, and so on. In other words, the KM contribution indicates the percentage of users who input answers without withholding answers to knowledge confirmation requests made by the system according to this embodiment.

[0124] In other words, when a knowledge confirmation is displayed to a user with a high KM contribution, there is a high possibility of obtaining a valid answer. In the process of Fig. 16, records in which the status information 133 is "suggestion" are excluded on the assumption that knowledge management thinking or operations are not particularly required of the user, but these may be included in the calculation of the KM contribution.

[0125] The plan effectiveness rate indicates the percentage of tasks that the user 80, who is the work planner, determined to be necessary in advance in the work plan were also necessary to proceed with the actual work. This is an index that indicates the accuracy of the work planner's practical work knowledge. The field suitability rate indicates the percentage of the work planner's specialty field that matches the specialty field related to each work condition.

[0126] The specialized field related to each work condition is determined by referencing the node ID of the fromID information 121 and toID information 122 of the work condition table in the work content table 11, extracting all the classification information 153 assigned to the word information 150 in the work word table 15 contained in the keyword information 112, and storing it in the field code of the work condition table 12. This is an index showing the accuracy of the theoretical work knowledge of the work planner.

[0127] As an example, the "weight" of a work planner can be the average value of KM contribution, plan effectiveness rate, and field suitability. When calculating the reliability of a work condition, first obtain from the reliability table 13 which of all work planners has affirmed the decision, then calculate the reliability using the following [Equation 1] and store it in the reliability information 124. Note that the function I is I(x)=1 if the condition x is satisfied, and I(x)=0 if the condition x is not satisfied. R is the reliability of the work condition. U is the "work planner" and Uqty is the "total number of work planners." J U is the judgment result for the work planner, and "J U "=True" indicates that the result of the determination is positive. U is the "weight" for the work planner.

[0128]

number

[0129] Furthermore, it is preferable to exclude work conditions that are rejected by many work planners from the targets of work suggestions by this system. Therefore, even if the judgment is negative, the reliability is calculated as in [Equation 1], and if the value is equal to or greater than a threshold value such as "0.2", an abolishment flag is added to the reliability information 124. As described above, the reliability calculation unit 41 and the change confirmation unit 73 can update the reliability information 124 so that the frequency with which the work conditions for which the answer received from the user is other than positive is reduced in the work condition search unit 42.

[0130] FIG. 17 is a flowchart of a work condition search routine executed in the work condition search unit 42. Through this routine, the work condition search unit 42 retrieves the work condition table 12 (see FIG. 3), searches for work conditions that satisfy the node of the fromID information 121 based on the occurring event input by the user 80 in the proposal display unit 71, and proposes the node of the toID information 122 to the user.

[0131] (S4202: Data entry) In Fig. 17, when the process proceeds to step S4202, work condition search unit 42 executes a data input process. That is, work condition search unit 42 acquires occurring event information through input from user 80. Here, the occurring event information is information including one or more pieces of keyword information 112 in which node type information 115 (see Fig. 2) is an "event", such as the model name in which an error occurred in maintenance target equipment 300, and an error code. Work condition search unit 42 acquires one or more pieces of node ID information 116 corresponding to the keyword information 112 that is this "event". Hereinafter, the acquired node ID information 116 will be referred to as a "query ID".

[0132] (S4204: Get work conditions) Next, when the process proceeds to step S4204, the work condition search unit 42 executes a work condition acquisition process, the details of which are as follows. (1) Create a “Proposed Work List” which is an empty set. (2) From the work condition table 12 (see FIG. 3), all records whose reliability information 124 is equal to or greater than a predetermined value are acquired. The collection of node ID information 120 of the acquired records is called an "acquired node ID list." In this way, by acquiring all records whose reliability information 124 is equal to or greater than a predetermined value,

[0133] (S4206: Proposed work search) Next, when the process proceeds to step S4204, the work condition search unit 42 executes a suggested work search process, the details of which are as follows. (1) Select one piece of node ID information 120 that is included in the acquired node ID list and has not been subjected to the process of (2) below. (2) For the selected node ID information 120, if the condition "the corresponding fromID information 121 is included in the query ID or the proposed work list" is satisfied, the toID information 122 corresponding to the node ID information 120 is added as an element to the proposed work list. Also, the node ID information 120 is set as the "work condition" corresponding to the element.

[0134] (S4208: Have all node ID lists been processed?) Next, when the process proceeds to step S4208, the work condition search unit 42 determines whether or not the process of step S4206 described above has been performed on all node ID information 120 included in the acquired node ID list. If the determination is "No", the process returns to step S4206. On the other hand, if the determination is "Yes", the process proceeds to step S4210.

[0135] (S4210: Work proposal) Next, when the process proceeds to step S4210, the work condition search unit 42 performs a work proposal process. That is, the work condition search unit 42 presents the contents of the proposed work list to the user 80 via the proposal display unit 71 (see FIG. 1).

[0136] In relation to the above processing, the work condition search unit 42 obtains the work condition table 12, extracts some work conditions from it, and uses the counterfactual verification unit 72 to request the user 80 to input judgment information 134 for those work conditions. In order to perform the reliability calculation unit 41 described above with high accuracy, it is preferable to obtain judgments of the work conditions from many work planners. Furthermore, it is also preferable to accelerate the accumulation of judgments regarding work conditions associated with events that occur infrequently.

[0137] FIG. 18 is a flowchart of a work condition testing routine executed in the work condition testing unit 43. (S4302: Data entry) In Fig. 18, when the process proceeds to step S4302, work condition testing unit 43 executes a data input process. That is, work condition testing unit 43 acquires occurring event information through input from user 80. Here, the occurring event information is the same as that described in step S4202 (see Fig. 17). Then, similar to the process in step S4202, one or more pieces of node ID information 116 corresponding to the occurring event information are acquired as a "query ID".

[0138] (S4304: Calculate judgment ratio) Next, when the process proceeds to step S4304, the work condition testing unit 43 executes a judgment ratio calculation process, the contents of which are as follows. (1) Obtain the proposed work list (created by the work condition search unit 42) displayed in the proposal display unit 71 (see FIG. 1). Also create an empty "test candidate list." (2) Compare the contents of fromID information 121 in each record of the work condition table 12 (see FIG. 3) with the query ID, and add the node IDs of rows that match partially but not completely to the test candidate list. Each element in the test candidate list is called a "test candidate ID." Each element in the proposed work list is called a "proposal ID."

[0139] (3) The node ID information 120 of a record in which the toID information 122 in the work condition table 12 (see FIG. 3) matches any of the proposed IDs is added to the test candidate list as a new test candidate ID. (4) Obtain each test candidate ID and the corresponding column of judgment information 134 in the reliability table 13 (see FIG. 4), and calculate the ratio of “positive” and “negative” judgment information 134 for each planner ID information 132.

[0140] (S4306: Update reliability table) Next, when the process proceeds to step S4304, the working condition testing unit 43 executes a reliability table update process, the details of which are as follows. (1) In the operation condition table 12 (see FIG. 3), a record including fromID information 121 corresponding to each test candidate ID is searched for. The node ID information 120 of the searched record is called a search target node ID.

[0141] In the work history table 22 (see FIG. 8), obtain the instruction ID column (contents of the planned work information 224) of the record that contains the search target node ID in the occurred event information 223. Obtain all instruction ID columns that correspond to the search target node ID from the work history table 22, and calculate the average rate (average rate) of the appearance of the search target node ID in these instruction ID columns.

[0142] (2) Calculate the “weight” of user 80 by the same process as that of reliability calculation unit 41. (3) The calculated average ratio and the "weight" are summed for each test candidate ID. Test candidate IDs whose summation result falls within a predetermined upper range and is equal to or greater than a predetermined threshold are extracted. The work conditions (contents of the work condition table 12 (see FIG. 3)) of the extracted test candidate ID are confirmed by the counterfactual verification unit 72. (4) The user 80 writes the result of affirming, denying, or reserving the work condition into the determination information 134 of the reliability table 13 (see FIG. 4).

[0143] 18 targets work conditions for which "the event input by the user 80 in the proposal display unit 71 or the work content proposed by the processing of the work condition search unit 42" corresponds to part of the fromID information 121 or toID information 122. The extracted work conditions do not correspond to the case targeted in the proposal display unit 71, but indicate that they correspond to a similar case. In this embodiment, confirming with the user 80 whether the work content in the toID information 122 is necessary if the fromID information 121 of the extracted work condition is hypothetically satisfied in that similar case is referred to as counterfactual recommendation.

[0144] It is necessary to calculate the priority order of the task conditions for executing the counterfactual recommendation from among the extracted task conditions. Specifically, priority is given to task conditions for which the reliability judgment is insufficient, the occurrence frequency of the event in the fromID information 121 is low, and there is a high possibility of obtaining a valid judgment from the user 80. To judge the insufficient reliability judgment, the proportion of planner IDs for which positive or negative is stored in the judgment information 134 in the reliability table 13 is calculated for each node ID.

[0145] Furthermore, to determine the rarity of an event's occurrence, the average rate at which each event in the fromID information 121 is included in the occurred event information 223 of the work history table 22 is calculated. The possibility of obtaining a valid judgment is calculated based on the work planner's weight in the reliability calculation unit 41. For example, the priority p of each work condition can be calculated by the following [Equation 2]. In [Equation 2], a, b, and c are weights. α represents the degree of "lack of reliable judgment", β represents the degree of "rarity of event occurrence", and γ represents the degree of "possibility of obtaining a valid judgment".

[0146]

number

[0147] The work conditions that are among the top few candidates with high priority p calculated by the above formula and whose priority p exceeds a threshold value can be confirmed by the counterfactual simulation confirmation unit 72. Therefore, when the user 80 confirms, denies, or reserves the candidate in question during confirmation by the counterfactual simulation confirmation unit 72, the work condition testing unit 43 stores the result in the reliability table 13 as determination information 134.

[0148] The work condition estimation unit 44 estimates likely work conditions having the work contents before and after the change in the toID for the work contents input by the user 80 in the proposal display unit 71, and stores them in the work condition table 12 after confirmation by the user 80 by the change confirmation unit 73. This process is preferably provided to correct errors in the existing knowledge of the system while referring to minimal judgment by the user 80. This process should be executed each time the user 80 makes a change to the work contents proposed to the user 80 in the proposal display unit 71.

[0149] Changes that user 80 can make are assumed to include adding, modifying, and deleting work content. For adding and modifying changes, existing node ID information 116 in work content table 11 or new original text information 113 is accepted as the new work content. When the latter new original text information 113 is input, a new row is added to work content table 11 through processing by work content extraction unit 31, and its node ID is obtained. For modifying and deleting changes, an existing node ID is accepted as the work content before the change. For adding changes, the node ID after the change is processed, for deleting changes, the node ID before the change is processed, and for modifying changes, the node IDs before and after the change are processed.

[0150] FIG. 19 is a flowchart of a work condition estimation routine executed by the work condition estimation unit 44. (S4402: Data entry) In FIG. 19, when the process proceeds to step S4402, the work condition estimation unit 44 executes a data input process. That is, the work condition estimation unit 44 acquires a query ID based on the occurring event information input by the user 80 in the proposal display unit 71. This process is similar to that of step S4202 (see FIG. 17) described above. In addition, the work condition estimation unit 44 acquires the ID of the changed work or work element changed by the user 80 in the proposal display unit 71 as a change ID.

[0151] (S4404: Update attention work table) Next, when the process proceeds to step S4404, the work condition estimation unit 44 executes the target work table update process according to the contents listed below. (1) Identify records in which the toID information 122 in the work condition table 12 (see FIG. 3) is a change ID. Then, obtain the number of node IDs included in the fromID information 121 in each of the identified records. Then, obtain the median value of the obtained "number of node IDs" as the "number of inference events."

[0152] (2) Enumerate all combinations of query IDs for the number of inference events. The set of enumerated combinations is called the “combination set.” (3) Among the combinations of query IDs listed in (2), obtain fromID information 121 of all records whose reliability information 124 is “abolished” in the work condition table 12 (see FIG. 3). Exclude from the combination set any combinations that are the same as the node ID combinations listed in the obtained fromID information 121.

[0153] (4) For any combination included in the combination set, extract a record from the work history table 22 that includes all elements included in the combination as the occurrence event information 223 (see FIG. 8). (5) The set of occurring event information 223 in the records extracted by (4) is called the "occurring event sequence," the set of case ID information 220 is called the "case ID sequence," and the set of valid task information 225 is called the "valid task sequence." It is determined whether the change ID found in step S4402 is included in each element of the valid task sequence. The set of this determination result is called the "task presence / absence sequence."

[0154] Then, the same number of records as the number of records extracted in (4) are added to the noted task table 24 (see FIG. 10). Then, the contents of the occurring event column, the case ID column, and the valid task column are written as the occurring event information 240, the case ID information 241, and the task presence / absence information 242 for these records.

[0155] (6) Records equal to the number of query IDs are added to the noted task table 24. These query IDs are written in the occurrence event information 240 of these records, and "Yes" is written in all task presence / absence information 242 of these records.

[0156] (S4406: Select unprocessed work condition candidates) Next, when the process proceeds to step S4406, work condition estimation unit 44 selects one record that has not been subjected to the process of step S4408 described later, from among the records in work condition table 12 (see FIG. 3). The selected record is called a work condition candidate.

[0157] (S4408: Percentage calculation) Next, when the process proceeds to step S4408, the work condition estimation unit 44 calculates the percentage of work condition candidates for which the work presence / absence information 242 is “present” when the conditions of the occurring event information 240 in the focus work table 24 (see FIG. 10) are satisfied.

[0158] (S4410: Have all work condition candidates been processed?) Next, when the process proceeds to step S4410, the work condition estimation unit 44 determines whether or not all records in the work condition table 12 have been processed as work condition candidates. If the determination here is "No", the process returns to step S4406. On the other hand, if the determination is "Yes", the process proceeds to step S4412.

[0159] (S4412: Display the work condition candidate with the highest ratio) Next, when the process proceeds to step S4412, work condition estimation unit 44 causes the work condition candidate having the maximum “ratio” calculated in step S4408 to be displayed in change confirmation unit 73. After checking the display content of change confirmation unit 73, user 80 inputs command information of “affirmative”, “negative”, or “withhold”.

[0160] (S4422: Setting judgment information) Next, when the process proceeds to step S4414, the work condition estimation unit 44 sets the judgment information 134 of the record related to the work condition candidate in the reliability table 13 (see FIG. 4) in accordance with the input command information. That is, if the input command information is "affirmative", "affirmative" is written into the judgment information 134, if the command information is "negative", "negative" is written into the judgment information 134, and if the command information is "pending", "pending" is written into the judgment information 134.

[0161] (S4424:Are there any edit or delete operations?) Next, when the process proceeds to step S4424, the work condition estimation unit 44 judges whether or not a change operation such as modification or deletion has been performed by the user 80. If the judgment here is "Yes", the process proceeds to step S4426. On the other hand, if the judgment here is "No", step S4426 is skipped.

[0162] (S4426: Set the working conditions before the change to negation) In step S4426, task condition estimation unit 44 writes "No" to determination information 134 in reliability table 13 (see FIG. 4) for the task condition before it was modified or deleted by user 80.

[0163] (S4428: Is change response processing required?) Next, when the process proceeds to step S4428, the work condition estimation unit 44 determines whether or not additional processing is required. That is, it determines whether or not there is a record in the work condition table 12 (see FIG. 3) in which the toID information 122 is the work before the change (hereinafter, referred to as a record requiring change). If the determination in step S4428 is "No," the processing of this routine ends.

[0164] (S4430:Resetting query ID and change ID) On the other hand, if the determination in step S4428 is "Yes", the process proceeds to step S4430. Here, the work condition estimation unit 44 sets the fromID information 121 of the change-required record to the above-mentioned query ID, and sets the ID of the work or work element before the change to the change ID. Next, the process returns to step S4404. As a result, the work condition estimation unit 44 repeats the processes of steps S4404 to S4428 based on the new query ID and the new change ID.

[0165] In this way, according to the process of FIG. When the work condition estimation unit 44 accepts a change, it obtains the node ID of the occurring event entered by the user 80 in the proposal display unit 71 and the node ID of the work content before and after the change as change IDs. The change ID is then searched for in the toID information 122 of the work condition table 12, and the median number of nodes whose node type in the fromID of the corresponding row is an event is calculated. The calculated value is then used as the number of events of the new work condition estimated in this process. However, if the likelihood of all work conditions is below an appropriate threshold in later processing, the positions may be added to the number of events and the process may be redone.

[0166] Combinations taken from the node IDs of the input occurring events are enumerated for the number of events determined here. However, when the change ID is the node ID before the change, the combinations are not enumerated for the node IDs of the input occurring events. That is, in this case, the change ID is searched for in the toID information 122 of the work condition table 12, and the fromID information 121 of the corresponding row is used to enumerate the combinations. Also, if the enumerated combination matches the fromID information 121 of the work condition table 12 and the reliability information 124 includes an abolition flag, the combination is excluded.

[0167] Next, for each combination listed, case ID information 220 is obtained that includes all of the occurring events of the combinations listed in the occurring event information 223 of the work history table 22. Then, for each case ID, whether or not the change ID is included in the planned work information 224 is tallied as work presence / absence information 242 in the noted work table 24. Then, the occurring event information 223 of the case ID information 220 is stored in the noted work table 24. Furthermore, the node ID of the entered occurring event can be stored in the noted work table 24 as occurring event information 240, and when the change ID is before the change, the work presence / absence information 242 can be set to "absent", and when the change ID is after the change, the work presence / absence information 242 can be set to "present", all of which are stored in the noted work table 24.

[0168] Here, reference will be made again to FIG. 10 to explain its contents. 10, the node IDs listed in the occurrence event information 240 correspond to the node IDs of the input occurrence events. Next, for each combination of the listed occurrence events, A. A combination of such occurrences; or B. Combination related to the event of fromID information 121 for a record in which "toID information 122 in the work condition table 12 is a change ID" When any of the above is included in the occurring event information 240 of the noted task table 24, the proportion of the task presence / absence information 242 that is "present" is calculated.

[0169] Then, the "enumerated combination of occurring events" with the highest ratio is confirmed by the change confirmation unit 73. However, if the node type of the change ID is a node with a higher node ID such as "repair" or "element", the higher node ID is added to the "enumerated combination of occurring events" and confirmed by the change confirmation unit 73.

[0170] If the confirmation by the change confirmation unit 73 confirms that the combination of occurrence events listed by the user 80 is positive, the determination information 134 is set as positive and stored in the reliability table 13. Alternatively, if the user 80 manually inputs correct task conditions, the task conditions are similarly stored in the reliability table 13. Furthermore, if the change made by the user 80 is a modification or deletion of task content, the task conditions used to recommend the task content before the change are stored in the reliability table 13 with the determination information 134 set as negative.

[0171] 20 and 21 are diagrams showing the flow of processing by the proposal display unit 71. FIG. The user 80 inputs an event that has occurred in a case for which an operation is to be performed into the event input section 710 (S7102, S7104).

[0172] Based on this input content, the proposal display unit 71 acquires a proposed work content through processing by the work condition search unit 42 of the work knowledge proposal unit 40 (S7106, S7108). Then, the proposal display unit 71 displays the acquired proposal content in the work content display unit 711 (S7110). The user 80 can input changes to the proposed work content, such as modification, addition, or deletion, by pressing the modification input button 712, the add work button 713, or the delete work button 714, respectively (S7112 to S7120).

[0173] It is now assumed that the change to the work content has been confirmed (S7130). The proposal display unit 71 executes the processing of the work condition estimation unit 44, and transitions the screen to the change confirmation unit 73 (S7132-S7146). In addition, at any time, such as when the processing of the work condition search unit 42 is completed or when the user 80 presses the plan completion button 716, the proposal display unit 71 executes the processing of the work condition testing unit 43, and transitions the screen to the counterfactual simulation confirmation unit 72 (S7112-S7120).

[0174] When the user inputs any work content in the work content display section 711, the work conditions used in the proposal are referenced from the work condition table 12 and displayed in the proposal basis display section 715 (S7130-S7146). After the user 80 has viewed all the work contents and made any appropriate changes, the user 80 presses the plan completion button 716, whereupon the work content selected by the user 80 is stored in the reliability table 13 and the processing of the proposal display section 71 ends (S7150).

[0175] FIG. 22 is a block diagram of a computer 980. The work knowledge storage unit 10, work history accumulation unit 20, work knowledge initialization unit 30, work knowledge proposal unit 40, and planner terminal 70 shown in Fig. 1 each include one or more computers 980 shown in Fig. 22. In Fig. 22, the computer 980 includes a CPU 981, a storage unit 982, a communication I / F (interface) 983, an input / output I / F 984, and a media I / F 985.

[0176] Here, the storage unit 982 includes a RAM 982a, a ROM 982b, and a HDD 982c. The communication I / F 983 is connected to a communication circuit 986. The input / output I / F 984 is connected to an input / output device 987. The media I / F 985 reads and writes data from a recording medium 988. The ROM 982b stores an IPL (Initial Program Loader) executed by the CPU, etc. The HDD 982c stores application programs, various data, etc. The CPU 981 executes application programs, etc. loaded from the HDD 982c to the RAM 982a to realize various functions.

[0177] The internal configurations of the work knowledge storage unit 10, the work history accumulation unit 20, the work knowledge initialization unit 30, the work knowledge proposal unit 40 and the planner terminal 70 shown in FIG. 1 are mainly shown as blocks of functions realized by application programs, etc.

[0178] [Effects of the embodiment] As described above, according to the embodiment, the operation knowledge proposal system 101 is an operation knowledge proposal system that proposes the necessary operation (K10, 121) and the operation element (K2, 122) including the parts or procedures related to the operation (K10, 121) when the need arises to perform the operation (K10, 121) on the maintenance target facility 300, and has a function of inputting and outputting data to and from the operation knowledge storage unit 10 that stores the operation (K10, 121), the operation element (K2, 122), and the operation condition (12R1, 12R2) that is the condition under which the operation (K10, 121) and the operation element (K2, 122) are required, and when the user 80 inputs any of the occurring event information 223, the operation knowledge proposal system 101 proposes the necessary operation (K10, 121) and the operation element (K2, 122) that correspond to the input occurring event information 223, and the operation knowledge proposal system 101 proposes the necessary operation (K10, 121) and the operation element (K2, 122) that correspond to the operation condition (12R1, 12R2). a proposal display unit 71 that displays the retrieved task conditions (12R1, 12R2), tasks (K10, 121) and task elements (K2, 122) and reliability information 124 corresponding to the task conditions (12R1, 12R2), and accepts changes from the user including addition, modification, or deletion of the tasks (K10, 121) or task elements (K2, 122); a task condition estimation unit 44 that, when a change is accepted, searches for and displays candidates for task conditions (12R1, 12R2) related to the tasks (K10, 121) and task elements (K2, 122) before or after the change based on a combination of the input occurring event information 223; and a change confirmation unit 73 that accepts an answer from the user 80 as to whether the candidates are valid or not, and adds the task conditions (12R1, 12R2) for which the answer is affirmative to the task knowledge storage unit 10. This makes it possible to provide the user with appropriate maintenance knowledge. For example, the reliability information 124 of the work conditions (12R1, 12R2) can be updated according to the action of the user 80 in response to the proposal of the work (K10, 121) or work element (K2, 122) or the confirmation of the work conditions (12R1, 12R2), so that appropriate maintenance knowledge can be provided to the user.

[0179] Moreover, the task knowledge proposal system 101 has a function of inputting and outputting data to and from the task knowledge storage unit 10 which further stores reliability information 124 of the task conditions (12R1, 12R2), and it is more preferable that the change confirmation unit 73 updates the reliability information so as to reduce the frequency with which task conditions (12R1, 12R2) for which the answer was other than affirmative are proposed in the task condition search unit 42. This can reduce the frequency with which task conditions (12R1, 12R2) with low reliability are proposed.

[0180] Moreover, it is more preferable that the task knowledge suggestion system 101 further includes a function of inputting and outputting data to and from the task history accumulation unit 20 that stores history (22R1, 22R2) including the task (K10, 121), occurrence date information 222 related to the task (K10, 121), and occurrence event information 223 representing an event corresponding to the task (K10, 121) or task element (K2, 122). This allows the task history accumulation unit 20 to be updated appropriately.

[0181] The work history accumulation unit 20 accumulates work content information 214 in which history (22R1, 22R2) relating to the executed work (K10, 121) and work element (K2, 122) and history (22R1, 22R2) of the occurrence of an event related to the occurring event information 223 are recorded in natural language. The work knowledge storage unit 10 has an work word table 15 including word information 150 indicating keywords that may appear in the work content information 214 and classification information 153 that defines the classification of the word information 150, and stores the work word table 15 for the work content information 214. It is even more preferable to further include a work knowledge initialization unit 30 that initializes the contents of the work knowledge memory unit 10, which has a work content extraction unit 31 that performs text analysis using the block 15 to extract tasks (K10, 121) and work elements (K2, 122), and a work condition extraction unit 32 that analyzes the extracted tasks (K10, 121), the structure in the work content information 214, and the relationship between the occurrence event information 223 in the multiple histories (22R1, 22R2), and the tasks (K10, 121), to estimate the work conditions (12R1, 12R2).

[0182] This allows the knowledge in the operation knowledge storage unit 10 to be properly initialized, so that more appropriate maintenance knowledge can be provided to the user.

[0183] The work history accumulation unit 20 records effective work information 225 indicating which of the planned works (K10, 121) or work elements (K2, 122) were actually effective, the work knowledge storage unit 10 records specialty field information 142 indicating the work field in which the user 80 specializes, the work conditions (12R1, 12R2) include work field information 123 indicating the field corresponding to the work (K10, 121) and the work elements (K2, 122), and the work knowledge storage unit 10 records the work content (130) of the work proposed to the user 80 according to the work conditions (12R1, 12R2), or the validity of the work conditions (12R1, 12R2). The reliability calculation unit 41 further includes a reliability calculation unit that assigns a weight to the user 80 based on the degree of agreement between the specialty information 142 and the work field information 123 and the frequency of the user 80's replies to the proposal or confirmation made to the user 80, and assigns higher reliability information 124 to the work conditions (12R1, 12R2) for which the user 80 has a higher rate of affirmation. It is more preferable that the proposal display unit 71 has a function to display the reliability information 124 of the work conditions (12R1, 12R2) used in the proposal and the work history by the user 80. As a result, the reliability calculation unit 41 assigns the reliability information 124 based on the result of the user 80's reply to the proposal or confirmation made to the user 80, so that more appropriate maintenance knowledge can be provided to the user.

[0184] Further, the work conditions (12R1, 12R2) correspond to a degree of infrequency of the related event K4, work (K10, 121) or work element (K2, 122) occurring or being performed in the work history storage unit 20, and the suggestion display unit 71 allows the user 80 to input a part of the event K4 related to the work conditions (12R1, 12R2) whose degree is equal to or greater than a predetermined value, or allows the user 80 to input a part of the work (K10, 121) or work element (K2, 122) related to the work conditions (12R1, 12R2) whose degree is equal to or greater than a predetermined value. 22) to the user 80, and further comprises a counterfactual scenario confirmation unit 72 that confirms with the user the validity of the work conditions (12R1, 12R2) that the related work (K10, 121) or work element (K2, 122) is necessary in a situation in which all events related to the work conditions (12R1, 12R2) whose degrees are equal to or greater than a predetermined value hypothetically occur, and it is more preferable that the work history accumulation unit 20 accumulates a history of responses by the user 80 to the confirmation in the counterfactual scenario confirmation unit 72. This makes it possible to respond to the history of responses by the user 80 in the counterfactual scenario confirmation unit 72, and therefore to provide the user with more appropriate maintenance knowledge.

[0185] Furthermore, when the user 80 inputs a part of the event K4 related to the work condition (12R1, 12R2) whose degree is equal to or greater than a predetermined value, or when the user 80 inputs a proposal for a work (K10, 121) or work element (K2, 122) related to the work condition (12R1, 12R2) whose degree is equal to or greater than a predetermined value, it is more preferable that the proposal display unit 71 increases the frequency of confirmation in the counterfactual scenario confirmation unit 72 for a combination of the user 80 and a work element (K2, 122) whose weighting given to the user 80 is equal to or greater than a predetermined value. This makes it possible to increase the frequency of confirmation in the counterfactual scenario confirmation unit 72 in accordance with the weighting given to the user 80, thereby making it possible to provide the user with even more appropriate maintenance knowledge.

[0186] [Variations] The present invention is not limited to the above-mentioned embodiment, and various modifications are possible. The above-mentioned embodiment is exemplified to explain the present invention in an easy-to-understand manner, and is not necessarily limited to those having all the configurations described. In addition, other configurations may be added to the configuration of the above-mentioned embodiment, and it is also possible to replace a part of the configuration with another configuration. In addition, the control lines and information lines shown in the figure show those that are considered necessary for explanation, and do not necessarily show all the control lines and information lines necessary on the product. In reality, it may be considered that almost all the configurations are connected to each other. Possible modifications of the above-mentioned embodiment are, for example, as follows.

[0187] (1) The hardware of the work knowledge storage unit 10, work history accumulation unit 20, work knowledge initialization unit 30, work knowledge proposal unit 40, and planner terminal 70 in the above embodiment can be realized by a general-purpose computer. Therefore, the processes corresponding to the above-mentioned block diagrams and flowcharts, as well as programs for executing the various processes described above, may be stored on a storage medium (a computer-readable recording medium on which a program is recorded) or distributed via a transmission line.

[0188] (2) In the above embodiment, the processes corresponding to the block diagrams and flowcharts, and the programs for executing the various processes described above have been described as software-based processes using programs. However, some or all of them may be replaced with hardware-based processes using an ASIC (Application Specific Integrated Circuit) or an FPGA (Field Programmable Gate Array), etc.

[0189] (3) The various processes executed in the above embodiments may be executed by a server computer (not shown) via network 60, and the various data stored in the above embodiments may also be stored in the server computer. [Explanation of symbols]

[0190] 10 Working knowledge storage section 11 Work Content Table 12 Working Condition Table 12R1, 12R2 Records (Working Conditions) 13 Confidence Table 13R1, 13R2 Records (Reliability History Information) 14 Work Planner Table 15 Working Word Table 20 Work history storage unit 21 Working Document Table 22 Work History Table 22R1,22R2 Records (History) 24 Featured Work Tables 30 Work knowledge initialization section 31 Work content extraction part 32 Work condition extraction unit 41 Reliability calculation unit 42 Work condition search section (work condition search process) 44 Working condition estimation unit (working condition estimation process) 71 Proposal display section (proposal display process) 72 Counterfactual Verification Section 73 Change confirmation section (change confirmation process) 80 users 101 Work knowledge suggestion system 121 fromID information (work) 122 toID information (work element) 123 Work Field Information 124 Reliability Information 130 Node ID information (work content) 134 Judgment information 142 Specialty Information 150 Word Information 153 Classification information 214 Work Content Information 222 Date of occurrence information 223 Incident Information 225 Valid Work Information 300 Maintenance Equipment 980 Computer K2 element (work element) K4 Event K10 Work MTC matrix V71 suggestion display screen V72 Counterfactual Confirmation Screen V73 Change confirmation screen

Claims

1. A work knowledge suggestion system that, when a need arises to perform work on a maintenance target facility, suggests a necessary work and a work element including a part or a procedure related to the work, A function of inputting and outputting data to and from a work knowledge storage unit that stores work, work elements, and work conditions that are conditions under which the work and the work elements are required; a task condition search unit that searches for task conditions corresponding to the inputted occurring event information and tasks and task elements required in accordance with the task conditions when the user inputs the occurring event information; a suggestion display unit that displays the searched task conditions, tasks, and task elements, as well as reliability information corresponding to the task conditions, and accepts changes from the user, including addition, modification, or deletion of tasks or task elements; a work condition estimation unit that searches for and displays candidates for work conditions related to the work and work elements before or after the change when the change is received based on the input combination of the occurring event information; a change confirmation unit that receives an answer from the user as to whether the candidate is valid or not, and adds the task condition for which the answer is affirmative to the task knowledge storage unit; An operation knowledge suggestion system comprising:

2. a function of inputting and outputting data to and from a task knowledge storage unit that further stores reliability information of task conditions; The change confirmation unit updates the reliability information so as to reduce the frequency with which the work condition search unit proposes the work condition for which the answer is other than affirmative.

2. The operation knowledge suggestion system according to claim 1 .

3. The system further includes a function of inputting and outputting data to and from a work history storage unit that stores a history including the work, occurrence date information related to the work, and occurrence event information indicating an event corresponding to the work or the work element.

3. The operation knowledge suggestion system according to claim 2.

4. the work history storage unit stores work content information in which the history relating to the performed work and the work elements and the history of an occurrence of an event relating to the occurrence event information are recorded in a natural language; the task knowledge storage unit includes a task word table including word information indicating keywords that may appear in the task content information and classification information defining classifications of the word information; a task content extraction unit that performs a sentence analysis on the task content information by using the task word table and extracts the task and the task element; and a task condition extraction unit that estimates task conditions by analyzing the extracted tasks, a structure in the task content information, and the relationship between the tasks and the occurrence event information in the multiple histories. The task knowledge initialization unit further includes a task knowledge storage unit that initializes the contents of the task knowledge storage unit.

4. The operation knowledge suggestion system according to claim 3.

5. the work history storage unit records effective work information indicating which of the works or work elements that were planned to be executed were actually effective; The work knowledge storage unit records specialty information indicating a work field in which the user specializes, The work conditions include work field information indicating a field corresponding to the work and the work element, the work knowledge storage unit accumulates reliability history information including determination information indicating a result of confirmation by the user of the work content of the work proposed to the user in accordance with the work conditions or the validity of the work conditions, a reliability calculation unit that assigns a weight to the user based on a degree of agreement between the specialty information and the work field information and a degree of frequency of the user's response to suggestions or confirmations made to the user, and assigns higher reliability information to the work conditions that are more frequently affirmed by the user; The proposal display unit has a function of displaying the reliability information of the work conditions used in the proposal and the work history of the user.

4. The operation knowledge suggestion system according to claim 3.

6. the task condition corresponds to a degree to which the relevant event, task, or task element occurs or is executed in the task history storage unit; the suggestion display unit allows the user to input a part of the event related to the work condition whose degree is equal to or greater than a predetermined value, or allows the user to input a suggestion for the work or the work element related to the work condition whose degree is equal to or greater than a predetermined value, a counterfactual confirmation unit that confirms to the user the validity of the work condition that the related work or work element is necessary in a situation where all events related to the work condition having the degree equal to or greater than a predetermined value have virtually occurred, The work history storage unit stores a history of responses by the user to confirmation by the counterfactual simulation confirmation unit.

4. The operation knowledge suggestion system according to claim 3.

7. When the user inputs a part of the event related to the work condition whose degree is a predetermined value or more, or when the user inputs a proposal for the work or the work element related to the work condition whose degree is a predetermined value or more, the suggestion display unit increases a frequency of performing the confirmation in the counterfactual scenario confirmation unit for a combination of the user and the work element whose weight assigned to the user is a predetermined value or more.

7. The operation knowledge suggestion system according to claim 6.

8. A work knowledge suggestion method for suggesting a necessary work and a work element including a part or a procedure related to the work when a work needs to be performed on a maintenance target facility, comprising: A process of inputting and outputting data to and from a work knowledge storage unit that stores work, work elements, and work conditions that are conditions under which the work and the work elements are required; a task condition search process for searching, when a user inputs any occurrence event information, for task conditions corresponding to the input occurrence event information and tasks and task elements required in accordance with the task conditions; a proposal display process for displaying the retrieved task conditions, tasks, and task elements, and reliability information corresponding to the task conditions, and accepting modifications, including addition, modification, or deletion of tasks or task elements, from the user; a task condition estimation process for searching for and displaying task condition candidates related to tasks and task elements before or after the change when the change is received based on the input combination of the occurring event information; a change confirmation process of receiving a response from the user as to whether the candidate is valid or not, and adding the task condition for which the response is affirmative to the task knowledge storage unit; A work knowledge suggestion method comprising:

Images