JP2018005554A - Fault tree generation device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a fault tree generation device that generates a fault tree of high precision from less fault information in the case of generating the fault tree from information on trouble occurring in an operation stage.SOLUTION: A fault tree generation device according to one embodiment of the present invention includes: an information narrowing-down unit which narrows down fault information to be used to generate a fault tree from among a plurality of pieces of fault information as instructed by a user; a fault tree generation unit which generates one or a plurality of fault trees from the fault information narrowed down by the information narrowing-down unit; a fault tree selection unit which allows the user to select a fault tree to be saved from among the one or plurality of generated fault trees; and a fault tree saving unit which saves the fault tree(s) selected by the user in a storage unit.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a fault tree creation apparatus.

  FTA (Fault Tree Analysis) is used as an analysis method for quality, reliability, maintainability, safety, etc., at the design and development stages of systems and equipment, as well as quality at each stage of manufacturing, inspection, and operation. Widely used for management, business improvement, accident countermeasures, etc.

  The FT diagram used in FTA (Fault Tree diagram; hereinafter referred to as “fault tree”) is a detailed description of the troubles that occur in the product or work to be analyzed in detail while detailing and disassembling. It is described in a tree shape while being detailed to a level where disassembly is not necessary. In the fault tree, the relationship between upper and lower events is expressed using causal relationships.

  Regarding the creation of such a fault tree, a method for automatically creating a fault tree has been proposed. For example, there are a method of creating a fault tree from a state transition model represented by events and transitions, and a method of creating using a continuous-time Markov chain. These methods are a model for automatically creating a fault tree from a state transition model described in a design document or an operation history.

  In addition to the above, a method of automatically extracting causal relationships using a sentence of a failure analysis document after the occurrence of a failure and creating a failure tree is conceivable. For example, there are a method (Non-Patent Document 1) for extracting from words indicating causal relations and a method using co-occurrence (Non-Patent Document 2), as in the paper of Girju et al. Further, there is a method (Patent Document 1) that extracts the causal relationship by limiting by event type and position information.

JP 2012-68860 A

Roxana Girju and Dan Moldvan “Minning Answers for Causal Questions”,. In Proceedings of the American Association for Artificial Intelligence (AAAI) -Spring Symposium, 15-25. Nobuyuki Omori, Akinori Mori: Examination of causal relations about products and parts from defect case documents, Proc. Of the 18th Annual Conference of the Association for Natural Language Processing, pp. 1192-1195, 2012.

  Railway vehicles have become more difficult to solve problems when a failure occurs than in the past, due to the increase in electronic parts that are procured by systematization in recent years. Also, unlike mass-produced products such as automobiles, it is difficult to perform sufficient test running in the sales section at the test stage and crushing the failure due to it, and it is necessary to implement step-by-step failure countermeasures and improvements even after the start of operation. It is very important to collect and analyze failure information after the start of operation, in addition to failure analysis.

  However, the number of rail cars produced is much smaller than that of automobiles. Furthermore, although the number of vehicles is small, the traveling distance is long and the traveling environment and conditions are various. Therefore, in the case of a railway vehicle, the contents of the troubles that occur are various and varied, the reproducibility of the problems is low, and it is difficult to collect information under the same conditions.

  On the other hand, conventional methods such as a method using co-occurrence are techniques for analyzing a failure using natural language processing using statistics, and require a lot of information to improve the accuracy of the analysis. . For this reason, if an attempt is made to create a failure tree using conventional technology for an object with a small amount of information that can be acquired, such as a railway vehicle, a failure tree that accurately represents the features of the failure may not be created. . When creating a fault tree in a field where the amount of fault information that can be acquired is small, it is necessary to use a new method in addition to the conventional technique.

  A fault tree creation device according to an embodiment of the present invention includes an information narrowing unit that narrows down fault information to be used for creating a fault tree from a plurality of pieces of fault information based on a user instruction, and an information narrowing unit. A fault tree creating unit that creates one or a plurality of fault trees from the narrowed down fault information.

  According to the present invention, it is possible to create a fault tree with high accuracy even when the amount of fault information that can be acquired is small.

10 is a flowchart illustrating a processing procedure of the fault tree creation apparatus according to the first embodiment. 1 is a block diagram illustrating an example of an internal configuration of a fault tree creation apparatus according to Embodiment 1; A figure showing an example of search conditions for refinement Diagram showing an example of the first half of failure related information Diagram showing an example of the second half of failure related information Diagram showing an example of a fault tree library A diagram showing an example of the dictionary 244 Example of a screen for entering search conditions to narrow down fault information when creating a fault tree Example screen for selecting a created fault tree The flowchart which shows the processing procedure when carrying out the creation of the fault tree using the expression showing the causal relationship Flow chart showing the processing procedure when creating a fault tree using bigram The flowchart which shows the processing procedure which extracts subject word, phenomenon and cause from the rule of every expression which shows causal relation FIG. 7 is a block diagram illustrating a configuration example of a fault tree creation system according to a second embodiment. Flowchart of processing for accommodating search conditions between fault tree creation devices Flowchart of processing for interchange of fault trees between fault tree creation devices FIG. 6 is a block diagram illustrating a configuration example of a fault tree creation system according to a third embodiment.

  Several embodiments will be described below with reference to the drawings.

  FIG. 1 is a flowchart illustrating an example of a fault tree creation processing procedure in the fault tree creation apparatus according to the first embodiment. FIG. 2 is a block diagram of an example of a fault tree creation apparatus according to the first embodiment. In the following, the configuration of the fault tree creating apparatus 20 will be described first with reference to FIG.

  The failure tree creation device 20 displays failure information, countermeasure information, an input unit 21 for inputting search conditions used when referring to them and when creating failure trees, and displays information reference and search results. An output unit 22 that performs failure phenomenon / cause extraction processing and failure diagnosis calculation processing, failure information used for creating a failure tree, search conditions for the created failure tree and failure tree, and the like. It has the memory | storage part 24 to preserve | save.

  The input unit 21 is an input device used by a user to input information, such as a keyboard and a mouse, and the output unit 22 is a display device such as a display. As an example, the storage unit 24 is a device including a nonvolatile storage medium such as a hard disk drive (HDD). As shown in FIG. 2, the storage unit 24 stores information such as a search condition 241 for narrowing down, failure-related information 242, a failure tree library 243, a dictionary 244, and the like. The contents of these information will be described later.

  The arithmetic processing unit 23 includes an input from the input unit 21, an output unit 22, a control unit 231 that controls the functional units 232 to 236 belonging to the arithmetic processing unit, a narrowing search condition input unit 232, an information narrowing unit 233, A fault tree creation unit 234, a fault tree selection unit 235, and a fault tree storage unit 236 are included.

  The control unit 231 passes the information input to the input unit 21 to the other functional units (232 to 236), and outputs output information from the functional units (232 to 236) to the output unit 22. In addition, a fault tree is created and saved by executing each functional unit (232 to 236).

  The search condition input unit 232 for narrowing down causes the output unit 22 to display an information input screen (for example, FIG. 8) for the user to input search conditions and the like, and information input by the user using the input unit 21 (search condition) Etc.) is transferred to the information narrowing-down unit 233. The information narrowing-down unit 233 narrows down the failure information stored in the failure-related information 242 using the narrowing-down search condition passed from the narrowing-down search condition input unit 232, and the narrowed down failure information becomes the failure tree creation unit 234. To pass.

  The fault tree creating unit 234 creates a fault tree from the narrowed down information, and performs processing to pass to the fault tree selecting unit 235. The failure tree selection unit 235 causes the user to select a failure tree to be saved among the created one or more failure trees. Specifically, the failure tree selection unit 235 displays the created one or more failure trees on the output unit 22 via the failure tree selection screen shown in FIG. 9, and among the displayed one or more failure trees. The fault tree selected by the user using the input unit 21 is passed to the fault tree storage unit 236. The fault tree storage unit 236 stores the fault tree passed from the fault tree selection unit 235 in the fault tree library 243.

  For example, the fault tree creation apparatus 20 may be a general-purpose computer such as a personal computer (hereinafter simply abbreviated as “PC”). As is well known, a PC has a memory such as a DRAM and a microprocessor. When a PC is used as the fault tree creation device 20, the memory and the microprocessor function as the arithmetic processing unit 23. The memory stores a program for operating the PC as a device provided with each functional unit such as the control unit 231 and the fault tree creation unit 234 described above by being executed by the microprocessor. Each process described below is realized by the microprocessor executing this program.

  Conversely, the fault tree creation device 20 may be configured with dedicated hardware instead of a general-purpose computer such as a PC. A part of or all of the control unit 231, the search condition input unit 232 for narrowing down, the information narrowing down unit 233, the fault tree creation unit 234, the fault tree selection unit 235, and the fault tree storage unit 236 are hardware such as FPGA and ASIC. It may be comprised. Moreover, the failure tree creation apparatus 20 may include a component other than that described above. For example, an interface device (such as a network interface controller (NIC)) for communicating with other computers may be provided.

  Next, management information included in the fault tree creation apparatus 20 according to the present embodiment will be described. In this embodiment, for example, a description will be mainly given of an example in which a fault tree is generated from a collection result of cases when a fault occurs in a railway vehicle operated by a railway company.

  4 and 5 show an example of the storage format of the failure related information 242. FIG. The failure-related information 242 is information in which, for example, when a failure occurs in a railway vehicle operated by a railway company, the content of the failure that has occurred and information related to the failure. If this information is an example of a failure occurring in a railway company, for example, a train driver, a route maintenance person, etc. will fill in this information. Information about one failure is recorded in each row (record) of the failure-related information 242. In this embodiment, this one line of information is called “failure information” or “failure report”. 4 shows the first half of the failure report included in the failure-related information 242, and FIG. 5 shows the second half of the failure report.

  The failure-related information 242 accumulates failure information by adding new failure information every time a failure (railway vehicle failure or the like) occurs. The failure information is added to the failure-related information 242 by the failure information entry person (such as a maintenance person on the route) inputting the failure information using the input unit 21. Alternatively, failure information created by another device (such as a PC) may be added to the failure-related information 242 via a communication interface or the like, thereby accumulating information in the failure-related information 242.

  Each failure report of the failure-related information 242 includes management information 2421 for managing the failure, weather information 2422 as environment information at the time of the failure, traveling information 2423 related to operation, vehicle information 2424, and details of the failure. Includes fault details 2425 involved.

  The management information 2421 includes an ID 24211 for uniquely identifying a failure report, a failure name 24212, a drafter name 24213, a date 24214, and the like. The weather information 2422 includes a weather name 24221 such as sunny and rain, a wind direction 24222, a wind force 24223, a precipitation 24224, a snow depth 24225, and the like.

  The traveling information 2423 includes a schedule name 24231 of the train in which the failure has occurred, an occurrence section 24232, a traveling state 24235 such as traveling / stopped, and a section section 24236 indicating a power source such as DC / AC / diesel. The occurrence section 24232 is information regarding the position of the train when the failure occurs. When a failure occurs between the stations, “occurrence section 24232” is stored in the occurrence section 24232, and the station (1) at that time In (24233) and station (2) (24234), the station name of the section where the failure occurred is recorded. When a failure occurs in a train that is stopped at a station, “on the station” is stored in the generation section 24232, and the name of the station that is stopped is recorded when a failure occurs in the station (1) (24233).

  The vehicle information 2424 is information on a train in which a failure has occurred, and includes a train name 24241, a train number 24242, a car number 24243, a traveling direction 24244, and the like.

  The failure details 2425 include a failure content 24251 indicating the content of the failure, image information 24252 for supplementing the failure content, an analysis result 24253, a delay time 24254, time series information 24255, and the like. The analysis result 24253 is a column for entering the cause of the failure that has occurred and the detailed content of the failure. The content to be entered in this column is described by the creator of the failure report (the drafter 24213) in natural language. The delay time 24254 stores a delay time when the train is delayed due to a failure, and the time-series information 24255 is a column for recording events occurring at and after the failure.

  In the present embodiment, the fault details 2425 are mainly used for creating the fault tree, and particularly the information described in the analysis result 24253 is used. However, other information, for example, information entered in the management information 2421 or the like, may be used for creating a fault tree.

  FIG. 8 shows the search condition input unit 232 for narrowing down as an output unit so that the user can input conditions for narrowing down failure information used for creating a failure tree among a plurality of pieces of failure information stored in the failure related information 242. 22 is an example of a screen displayed on the screen. The search screen includes an accident occurrence date / time input field 71, a drafter input field 72, a weather information input field 73, a travel information input field 74, and a vehicle information input field 75.

  The accident occurrence date / time input column 71 is used to narrow down failure information by date 24214 in the management information 2421 in the failure related information 242. Similarly, in the drafter input field 72, the weather information input field 73, the travel information input field 74, and the vehicle information input field 75 are weather information 2422 and travel information 2423, respectively, in order to narrow down the failure information by the drafter 24213. The vehicle information 2424 is used to narrow down the failure information. When the user presses the search button 77 after inputting the information input from 71 to 75, the information narrowing unit 233 extracts information that matches the search condition from the information stored in the failure related information 242. .

  If the user wants to narrow down the failure-related information 242 using the past history (search conditions used in the past), the user selects the search history use check box 76 and presses the search button 77 to execute.

  Note that the search condition may be input to only one of these input fields, or the search condition may be input to a plurality of input fields. When search conditions are input in a plurality of input fields, the search condition input unit 232 for narrowing down extracts failure information that matches all of the input search conditions.

  The weather information input field 73 includes a weather name check box 731, a wind direction check box 732, a wind force input field 733, a precipitation amount input field 734, and a snow depth input field 735, each of which is a weather 24221 in the failure related information 242. , Wind direction 24222, wind force 24223, precipitation 24224, and snow depth 24225.

  The travel information input field 74 includes a diagram information input field 741, an occurrence section input field 742, a travel status check box 743, and a section classification check box 744. The diagram information input column 741 is used for narrowing down the diamond 24231 in the failure related information 242, and the generation section input column 742 narrows down the generation section 24222, the station 1 (24233), and the station 2 (24234). The travel situation check box 743 and the section section check box 744 are used to narrow down the travel situation 24235 and the section section 24236, respectively.

  The vehicle information input field 75 includes a train name field 751, a train number field 752, a car number 753 field, and a traveling direction check box 754 made up and down, and these are information items in the vehicle information 2424 (trains Name 24241, train number 24242, car number 24243, traveling direction 24244).

  The information included in the generated fault tree is information existing in the fault details 2425, particularly the analysis result 24253. As shown in FIG. 8, the fault information is narrowed down (search condition specified by the user). Is not the failure details 2425 (analysis result 24253) itself but the environment information / operation information at the time of occurrence of the failure. Specifically, the environment information / operation information includes weather information 2422, travel information 2423, and vehicle information 2424. Although these are not information describing the direct cause and contents of the failure, the cause of the failure of the railway vehicle often depends on the weather and the environment such as the operation section. Therefore, if the user can narrow down the failure information used for creating the failure tree based on the weather information 2422 and the traveling information 2423, the possibility that a highly accurate (likely) failure tree is created increases.

  FIG. 3 shows an example of the storage format of the search condition 241 for narrowing down. The search condition 241 for narrowing down stores the search condition input by the user from the search screen shown in FIG. 8 (to be exact, the generation of a fault tree that the user has determined to be appropriate among the search conditions input by the user) The search criteria used at times are saved). Therefore, as shown in FIG. 3, the search condition 241 for narrowing down includes columns for storing weather information, travel information, and vehicle information. The search condition input by the user from the search screen of FIG. It can be stored. Since the information stored in each column of the search condition 241 for narrowing down is the same as the information included in the search condition described with reference to FIG. 8, the description thereof is omitted here.

  FIG. 6 shows an example of the storage format of the fault tree library 243.

  The fault tree is composed of a top event, a plurality of intermediate events, and a basic event causing the fault tree, and the fault tree is stored in each row (record) in the fault tree library 243. The fault tree creation apparatus 20 according to the present embodiment attaches a unique identification number to each generated fault tree and stores it in the fault tree library 243. This identification number is called ID, and the ID is stored in the column 2431 of the fault tree library 243. The top event 2432, the intermediate event 2433, and the basic event 2434 store the top event, the intermediate event, and the basic event of the fault tree, respectively. For example, the failure point is stored in the top event, the phenomenon is stored in the intermediate event, and the cause of the failure is stored in the basic event.

  In the present embodiment, in a fault tree creation process described later, a process in which an intermediate event generates one fault tree, that is, a fault tree including one top event, one intermediate event, and one basic event is generated. The process will be mainly described. However, the fault tree creation process itself can be realized by using various known methods, and a process for creating a fault tree having a plurality of intermediate events may be executed by a fault tree creation process described later.

  In this embodiment, the fault tree is stored in a table like the fault tree library 243 shown in FIG. Therefore, for example, when there are a plurality (for example, two) of fault trees having a common top event, a plurality of rows in which the same information is stored in the top event 2432 are created. For example, in the example of FIG. 6, there are two rows (failure trees) whose top event 2432 is “monitor” (failure trees with IDs 2431 of 1 and 2). However, the data structure for storing the fault tree is not limited to the table structure. For example, a fault tree may be expressed using a list structure. In that case, a structure in which the top event and the intermediate event are shared by a plurality of fault trees may be adopted.

  FIG. 7 shows an example of the storage format of the dictionary 244. Each line of the dictionary 244 stores information on a subject word, a phenomenon, and a position where the cause is described for each word or expression indicating a causal relationship. This information is called a “rule” in this embodiment. How the dictionary 244 is used will be described later.

  FIG. 9 is an example of a failure tree selection screen displayed on the output unit 22 in order for the failure tree selection unit 235 to cause the user to select a failure tree. It comprises a check box 82 for allowing the user to select a fault tree that accurately expresses the contents of the fault among the fault trees.

  Of the lines displayed on the failure tree selection screen, the user checks the check box 82 of the line where the failure tree to be saved is displayed, and presses an OK button 83. When the OK button 83 is pressed, the failure tree selection unit 235 selects the failure tree in the row in which the check box 82 is checked, and the selected failure tree is stored in the failure tree library 243. At the same time, the search condition when the fault tree selected here is generated is stored in the search condition 241 for narrowing down in the storage unit 24.

  On the other hand, when the Cancel button 84 is pressed, none of the fault trees displayed on the fault tree selection screen is stored in the fault tree library.

  From here, the details of the fault tree creation processing performed by the fault tree creation apparatus 20 will be described. First, the overall flow of the fault tree creation process will be described with reference to FIG. When the user instructs the fault tree creation device 20 using the input unit 21, for example, the fault tree creation process is started.

  Step S102: The narrow-down search condition input unit 232 displays the failure information search condition input screen shown in FIG. The user uses this search condition input screen to input operation conditions for narrowing down failure information. When the condition input by the user is completed and the user presses the search button 77, the process proceeds to step S103.

  Step S103: The arithmetic processing unit 23 of the fault tree creation system 20 reads all the fault information included in the fault-related information 242 of the storage unit 24 into the arithmetic processing unit 23, for example, a memory ( Temporarily save to (not shown).

  Step S104: The information narrowing-down unit 233 narrows down to a plurality of pieces of failure information read in Step 103 that match the condition input by the user in Step S102. Here, failure information that does not match the conditions entered by the user is discarded from the memory, and the failure information remaining on the memory becomes the narrowed failure information.

  Step S105: The arithmetic processing unit 23 starts a loop (step S105 to step S107) for creating a fault tree for each piece of fault information narrowed down in step S104. In step S105, one of the failure information narrowed down in step S104 is selected, and the processing in steps S106 and S107 is executed for the selected failure information. Steps S106 and S107 are executed by the fault tree creation unit 234.

  Step S106: The fault tree creation unit 234 creates a fault tree from the fault information selected in step S105, and then proceeds to step S107. Although any known technique may be used for the process of creating the fault tree, an example of the technique will be described later with reference to FIGS.

  Step S107: The failure tree creation unit 234 temporarily stores the failure information search condition used in step S104 and the created failure tree in the memory of the arithmetic processing unit 23, and proceeds to step S108.

  Step S108: The arithmetic processing unit 23 determines whether the creation of the fault tree is completed for all the fault information narrowed down in step S104. If the creation of the fault tree has been completed, step S109 is performed next. If failure information that has not been processed in step S106 remains, the process returns to step S105.

  Step S109: The failure tree selection unit 235 displays the failure tree temporarily stored in step S107 on the screen in the format shown in FIG. 9, and proceeds to step S110.

  Step S110: The failure tree selection unit 235 causes the user to select a failure tree that is determined to be appropriate by the user from the failure trees displayed on the failure tree selection screen in FIG. The selection method is as described above, and the user checks the check box 82 of the row displaying the failure tree to be saved among the rows displayed on the failure tree selection screen, and presses the OK button 83. The fault tree is selected. After the selection information input is completed, the process proceeds to step S111. However, if the user determines that none of the one or more fault trees displayed on the fault tree selection screen is valid and presses the Cancel button 84, the fault tree selection unit 235 creates the memory in the memory created in step S106. The temporarily stored fault tree is discarded, and the process is performed again from step S102. That is, the failure information search condition input screen is displayed on the output unit 22 again, and the user resets the search conditions.

  Step S111: The fault tree storage unit 236 stores the search condition used when creating the fault tree selected in step S110 in the search condition 241 for narrowing down in the storage unit 24, and the fault selected in step S110. The tree is stored in the fault tree library 243. After saving the fault tree and search conditions, the process ends.

  As described above, the failure tree creation apparatus 20 according to the present embodiment causes the user to narrow down the items to be used for creating the failure tree from among a plurality of pieces of failure information, and creates the failure tree using the narrowed failure information. . When narrowing down the failure information, for example, when the failure information relates to a railway vehicle, the user can narrow down the failure information based on information such as the weather and the traveling section of the vehicle. In many cases, the failure of a railway vehicle is related to external factors such as the weather and traveling section. Therefore, it is expected that a failure tree with high accuracy (high validity) can be created by narrowing down failure information used for creating a failure tree in advance using information such as weather and traveling section.

  Moreover, there is not much information that can be used for creating a fault tree among the information included in the fault information. For example, in the case of the failure-related information 242 (FIGS. 4 and 5) described in the present embodiment, only the analysis result 24253 is information used for creating a failure tree. In the fault tree creation apparatus 20 according to the present embodiment, the user can repeat a number of processes for creating and displaying a fault tree by inputting a search condition, so that a fault tree with high validity can be created from a small amount of information. I can expect.

  Note that the user may use the search conditions stored in the search condition 241 for narrowing down to narrow down. In step S102, when the user checks the search history use check box 76 and presses the search button 77, the search condition stored in the search condition for search 241 is reflected on the search screen of FIG. Subsequent processing is the same as that described above, and by performing the processing from step S103 onward, it is possible to generate a fault tree and store the fault tree.

  When a plurality of search conditions are stored in the search condition 241 for narrowing down, when the user checks the search history use check box 76 and presses the search button 77, the search unit 77 selects a plurality of search conditions. A screen, such as the table of FIG. 3, is displayed. The user selects one of the displayed search conditions, and the selected search condition is reflected on the search screen of FIG. Subsequent processing is the same as described above.

  In the search condition 241 for narrowing down, the search condition used when the fault tree determined to be appropriate by the user is created. Therefore, if the search condition stored in the search condition 241 for narrowing down is used, there is a high possibility that a fault tree with high validity will be created, and the validity is more efficient than when the user inputs the search condition from scratch. High fault tree can be created.

  The above is the fault tree creation process performed by the fault tree creation apparatus 20 according to the first embodiment. In the following, some examples of fault tree creation processing (step S106 in FIG. 1) performed by the fault tree creation apparatus 20 will be described. Note that when the fault tree creating apparatus 20 according to the present embodiment creates a fault tree from the narrowed down fault information, any known technique may be used, and the processing described here is necessarily performed. There is no.

  The flowchart in FIG. 10 represents a first example of the flow of processing performed in step S106 in FIG. In the process of FIG. 10, the fault tree creation unit 234 finds an expression indicating the cause and effect from the fault information narrowed down in step 105, and creates a fault tree based on the expression. Causal expressions include “cause”, “read to”, “bring”, “about”, “generate”, etc. in the case of verbs, and “because of, due to, because, as, since, for, so, etc.” ). These pieces of information are stored in the dictionary 244 in the storage unit 24 in the format shown in FIG.

  Step S902: The fault tree creation unit 234 performs morphological analysis on the input fault information to divide the sentence into words, and put part of speech for each divided word. In the present embodiment, an example will be described in which the fault tree creation unit 234 creates a fault tree by analyzing a sentence described in the analysis result 24253 among the input fault information. However, as another embodiment, other information such as the failure name 24212 and the failure content 24251 may be used when creating the failure tree.

  Step S903: The fault tree creation unit 234 selects one of the words decomposed in Step 902, and performs the processing after Step S904.

  Step S904: The fault tree creation unit 234 refers to the dictionary 244 to determine whether or not the selected word is an expression representing a cause and effect. If the expression represents a cause and effect, the process proceeds to step S905. If the expression is different, the process returns to step S903 to process the next word.

  Step S905: The fault tree creation unit 234 extracts a subject word, a phenomenon, and a cause based on the selected word. Details of this processing will be described later (FIG. 12).

  In step S906, the fault tree creation unit 234 creates a fault tree by combining the subject word-phenomenon-cause extracted in step S905. Then, the failure tree creation unit 234 stores the created failure tree in the failure tree library 243 and ends the process.

  The flowchart in FIG. 11 represents another example of the flow of processing performed in step S106 in FIG. FIG. 11 shows an example in which a fault tree is generated from the fault information narrowed down in step 105 using a co-occurrence expression (bigram).

  Step S1002: This process is the same as step 902, and therefore the description thereof is omitted here.

  Step S1003: The fault tree creation unit 234 creates a combination (bigram) of two words of a word necessary for analysis, such as a verb and a noun, and counts the number of appearances of each created bigram.

  Step S1004: The fault tree creation unit 234 extracts a set of two bigrams including the same word (referred to as a common word) from the created bigrams. For example, when there exists a bigram 1 consisting of the words A and B and a bigram 2 consisting of the words B and C, the bigram 1 and the bigram 2 include the common word B. Therefore, in this case, the fault tree creation unit 234 joins bigram 1 and bigram 2 with the common word B, creates a fault tree (fault tree consisting of words A, B, and C), and ends the process.

  The flowchart of FIG. 12 shows an example of the process performed in step S905 of the flowchart of FIG. 10, and extracts the subject word, the phenomenon, and the cause from the sentence using the information of the dictionary 244.

  The dictionary 244 stores in which position the subject word, phenomenon, and cause are described for each word or expression indicating a causal relationship. For example, if the word that indicates the cause-and-effect relationship is “due to” (...), the subject of the phrase and clause before the due to is the subject, the case where there is a complement, the phenomenon immediately after the due to Information such as being registered is registered. Hereinafter, an example of processing using the above rule will be shown. Note that information such as words indicating the causal relationship stored in the dictionary 244 in FIG. 7 is English. However, if the sentence to be analyzed is Japanese, information on the word indicating the causal relationship in Japanese is provided. A stored dictionary 244 is used.

  Step S1102: The fault tree creation unit 234 obtains a rule relating to a word indicating a cause and effect included in the input sentence from the dictionary 244, and proceeds to step S1103.

  Step S1103: The fault tree creation unit 234 applies a rule for extracting a subject word, and extracts a subject word. For example, in the case of due to, the subject of all due to phrases and clauses is extracted. After the extraction is completed, the process proceeds to step S1104.

  Step S1104: The fault tree creation unit 234 applies a rule for extracting a phenomenon to extract the phenomenon. For example, in the case of due to, a complement is extracted from the phrase and clause immediately before due to. After the extraction is completed, the process proceeds to step S1105.

  Step S1105: The fault tree creation unit 234 applies a rule for extracting the cause, and extracts the cause. For example, in the case of due to, the cause is extracted from a noun (noun phrase) immediately after due to. After extraction is complete, the process ends.

  In the second embodiment, a description will be given of an example in which a search condition for narrowing down and a failure tree countermeasure library are interchanged among a plurality of failure tree creation apparatuses. For example, if each department in a company has its own fault tree creation device and a fault tree is created using fault information collected independently by each department, a lot of fault information cannot be collected. May not be created. Therefore, in the second embodiment, by providing a plurality of failure tree creation devices with a function of mutually sharing failure trees and search conditions, a failure that has not been handled by each organization (department) having the failure tree creation device. An example will be described in which a fault tree can be used, or even when a single department does not have enough information to create a fault tree. A configuration example is shown in FIG.

  As illustrated in FIG. 13, the fault tree creation system according to the second embodiment has a configuration in which a fault tree creation device 120 and a fault tree creation device 121 are connected via a network 30. The failure tree creation device 120 and the failure tree creation device 121 may be present in the same place, for example, those owned by the same organization or department, but here each device is a different organization or An example of a department will be described. The failure tree creation device 120 and the failure tree creation device 121 may both be owned by the same company, or the company having the failure tree creation device 120 and the company having the failure tree creation device 121 are different companies. Further, a configuration in which a fault tree and a search condition are interchanged between companies may be used.

  Each of the fault tree creation device 120 and the fault tree creation device 121 is the same as that described in the first embodiment, the input units 1201 and 1211, the output units 1202 and 1212, the arithmetic processing units 1203 and 1213, and the storage unit 1204. 1214. Furthermore, the fault tree creating apparatus 120 and the fault tree creating apparatus 121 according to the second embodiment include communication units 1205 and 1215.

  The flowchart of FIG. 14 is an example of accommodating search conditions for narrowing down failure information when a plurality of failure tree creation apparatuses shown in FIG. 13 are connected via a network. Hereinafter, the side (accommodation destination) that receives the search condition accommodation will be described as a failure tree creation apparatus 120, and the side that provides the search condition accommodation (accommodation source) will be described as the failure tree creation apparatus 121.

  When the user requests the interchange processing of search conditions from the input unit 1201 at the fault tree creating apparatus 120, the fault tree creating apparatus 120 starts the process of FIG.

  Step S1302: The arithmetic processing unit 1203 requests the fault tree creation apparatus 121 via the communication unit 1205 for interchange processing of search conditions, and proceeds to step S1303.

  Step S1303: Upon receiving a request from the accommodation source via the communication unit 1215, the fault tree creating apparatus 121 reads the search condition from the storage unit 1214 in the arithmetic processing unit 1213, and the read search condition is transmitted to the communication unit 1215. To the fault tree creating apparatus 120 via The fault tree creation apparatus 120 stores the received search condition in the internal memory of the arithmetic processing unit 1203, and the process proceeds to step S1304.

  Step S1304: The arithmetic processing unit 1203 starts loop processing for all the search conditions received from the fault tree creating apparatus 121. Here, the arithmetic processing unit 1203 selects one of the received search conditions of the fault tree creating apparatus 121, and performs the processing after step S1305. In the following description, the search condition selected here is referred to as “accommodation source search condition”.

  Step S1305: The arithmetic processing unit 1203 selects one of the search conditions stored in the fault tree creation device 120, and proceeds to step S1306. The search condition selected here is called “accommodation destination search condition”.

  Step S1306: The arithmetic processing unit 1203 compares the accommodation source search condition selected in steps S1304 and S1304 with the accommodation destination search condition. If the search conditions are the same, the process proceeds to step S1307. If the search conditions are different, the process proceeds to step S1308.

  Step S1307: The arithmetic processing unit 1203 discards the accommodation source search condition, and proceeds to step S1308.

  Step S1308: If all the search conditions stored in the fault tree creation device 120 have been determined, the process proceeds to step S1309. If there are any search conditions that have not yet been determined, the process returns to step S1305.

  Step S1309: If all the search conditions received from the fault tree creating apparatus 121 have been determined, the process proceeds to step S1301. If there is a search condition that has not been determined among the received search conditions, the process returns to step S1304.

  Step S1310: The arithmetic processing unit 1203 stores the search conditions that are not discarded among the search conditions received from the fault tree creating apparatus 121 in the search conditions 241 for narrowing down in the storage unit 1204, and ends the process.

  The flowchart of FIG. 15 is an example in which a failure tree used for failure analysis is accommodated when a plurality of failure tree creation apparatuses shown in FIG. 13 are connected via a network. Hereinafter, the side (accommodation destination) that receives the search condition accommodation will be described as a failure tree creation apparatus 120, and the side that provides the search condition accommodation (accommodation source) will be described as the failure tree creation apparatus 121.

  When the user requests fault tree interchange processing from the input unit 1201 in the fault tree creating apparatus 120, the processing of FIG. 15 is started.

  Step S1402: The arithmetic processing unit 1203 requests the fault tree creating apparatus 121 via the communication unit 1205 for the fault tree interchange process, and the process proceeds to step S1403.

  Step S1403: Upon receiving a request from the accommodation source via the communication unit 1215, the failure tree creation device 121 reads out the failure tree from the storage unit 1214 in the arithmetic processing unit 1213, and transmits the read failure tree to the communication unit 1215. To the fault tree creating apparatus 120 via The fault tree creating apparatus 120 stores the fault tree accommodated by the fault tree creating apparatus 121 in the internal memory of the arithmetic processing unit 1203, and the process proceeds to step S1404.

  Step S <b> 1404: The arithmetic processing unit 1203 starts loop processing for all the fault trees of the fault tree creating apparatus 121 accommodated by the fault tree creating apparatus 121. Here, the arithmetic processing unit 1203 selects one of the fault trees received from the fault tree creating apparatus 121, and the process proceeds to step S1405.

  Step S1405: The arithmetic processing unit 1203 acquires the number of stages of the fault tree selected in step S1404, and proceeds to step S1406.

  Step S1406: The arithmetic processing unit 1203 repeats the processes of steps S1407 to S1413 for each fault tree stored in the fault tree creation apparatus 120. In step S1406, the arithmetic processing unit 1203 selects one of the fault trees stored in the fault tree creating apparatus 120.

  Step S1407: The arithmetic processing unit 1203 obtains the number of stages of the fault tree called in step S1406, and proceeds to step S1408.

  Step S1408: The arithmetic processing unit 1203 compares whether the top event of the fault tree called in steps S1404 and S1407 is the same. If the contents are the same, the process proceeds to step S1409; otherwise, the process proceeds to step S1413.

  Step S1409: The arithmetic processing unit 1203 sets the number of fault trees that have been accommodated as the upper limit of the number of processes after step S1410.

  Step S1410: The arithmetic processing unit 1203 starts a loop that repeats the processing in steps S1411 to S1413 for the number of stages of the fault tree called in step S1404, and proceeds to step S1411.

  Step S1411: The arithmetic processing unit 1203 compares the event of the accommodated fault tree with the content of the stored fault tree event. If they are the same, the process proceeds to step S1412, and if they are different, the process proceeds to step S1413.

  Step S1412: The arithmetic processing unit 1203 determines whether or not the subordinate event of the accommodated fault tree still remains. If it remains, the process returns to step S1410, and if not, the process proceeds to step S1414.

  Step S1413: The arithmetic processing unit 1203 stores the accommodated fault tree in the fault tree library 243 of the storage unit 1204, and the process proceeds to step S1415.

  Step S1414: The arithmetic processing unit 1203 determines whether there is any remaining fault tree that has been saved, and if there is, returns to step S1406. If not, the process proceeds to step S1415.

  Step S1415: The arithmetic processing unit 1203 determines whether there is a remaining in the accommodated fault tree. If there is, the process returns to step S1404, and if not, the process ends.

  The above is the description of the fault tree creation system according to the second embodiment. In the fault tree creation system according to the second embodiment, search conditions and fault trees can be interchanged between a plurality of fault tree creation apparatuses. In the above description, an example is described in which the fault tree creation apparatus 120 is an apparatus (accommodation destination) that receives search conditions and failure trees, and the fault tree creation apparatus 121 is an apparatus (accommodation source) that accommodates search conditions and fault trees. However, the failure tree creation device 120 may accommodate the search conditions and the failure tree to the failure tree creation device 121. In addition, here, an example in which two fault tree creation devices are included in the fault tree creation system has been described. However, even when three or more fault tree creation devices are included, the search described above between each fault tree creation process. Condition or fault tree interchange processing can be performed.

  In the processing described above (FIG. 15), both the fault tree creating apparatus 120 and the fault tree creating apparatus 121 hold the fault tree in the storage unit, and among the fault trees included in the fault tree creating apparatus 121, The example in which a fault tree (entire or part of the fault tree) that is not held by the fault tree creation apparatus 120 is added to the fault tree creation apparatus 120 has been described. However, when no fault tree is held in the storage unit of the fault tree creation apparatus 120, for example, the fault tree creation process may have never been performed in the fault tree creation apparatus 120. In that case, the fault tree creating apparatus 120 does not need to perform the processing of FIG. 15. After receiving all the fault trees included in the fault tree creating apparatus 121, the fault tree creating apparatus 120 stores all the received fault trees as they are. It may be stored in the part.

  The same processing may be performed when no search condition is held in the storage unit of the fault tree creating apparatus 120. That is, the failure tree creation device 120 may receive all the search conditions of the failure tree creation device 121 and store all the received search conditions in the storage unit of the failure tree creation device 120 as they are.

  In the above, an example in which search conditions and failure trees are interchanged between a plurality of failure tree creation devices has been described. However, information other than search conditions and failure trees may be interchanged between a plurality of failure tree creation devices. . For example, information stored in the dictionary 244 may be interchanged.

  FIG. 16 is a configuration example of a fault tree creation system according to the third embodiment. The fault tree creation system according to the third embodiment includes fault tree creation devices 120 ′ and 121 ′. Both of the fault tree creation apparatuses 120 ′ and 121 ′ have the same functions as the fault tree creation apparatus 20 described in the first embodiment. That is, the failure tree creation devices 120 ′ and 121 ′ all include a control unit 231, a search condition input unit 232 for narrowing down, an information narrowing down unit 233, and a fault tree creation unit in the arithmetic processing units (1203 ′ and 1213 ′). 234, a failure tree selection unit 235, and a failure tree storage unit 236.

  The configuration of the fault tree creation system according to the third embodiment is similar to that of the fault tree creation system according to the second embodiment. However, the fault tree creation apparatuses 120 and 121 described in the second embodiment have the storage units 1204 and 1214, respectively, whereas the fault tree creation apparatuses 120 ′ and 121 ′ according to the third embodiment have one There is a difference that the storage unit 24 'is configured to be shared.

  Both of the fault tree creation apparatuses 120 ′ and 121 ′ store the generated fault tree and the search conditions used when creating the fault tree in the storage unit 24 ′. Then, the fault tree creation device 121 ′ may refer to the search conditions and fault trees stored in the storage unit 24 ′ by the fault tree creation device 120 ′. Conversely, the fault tree creation device 120 ′ may refer to search conditions and fault trees stored in the storage unit 24 ′ by the fault tree creation device 121 ′. Therefore, unlike the fault tree creation system according to the second embodiment, the fault tree creation apparatuses 120 ′ and 121 ′ do not need to perform processing for transferring fault trees and search conditions to each other.

  In addition, the storage unit 24 ′ may be connected to the fault tree creation device 120 ′ or 121 ′ via a transmission line different from the network 30 as shown in FIG. 16, but as another embodiment, the storage unit 24 ′ (For example, a file server or the like) may be connected, and this storage device may be used as a storage unit shared by the fault tree creation devices 120 ′ and 121 ′. Although an example in which two fault tree creation devices are included in the fault tree creation system has been described here, in a fault tree creation system having three or more fault tree creation devices, each fault tree creation process stores one storage unit. It may be configured to share.

20: Fault tree creation device, 21: Input unit, 22: Output unit, 23: Arithmetic processing unit, 24: Storage unit, 231: Control unit, 232: Search condition input unit for refinement, 233: Information refinement unit, 234: Fault tree creation unit, 235: Fault tree selection unit, 236: Fault tree storage unit, 241: Search condition for narrowing down, 242: Fault related information, 243: Fault tree library, 244: Dictionary

Claims (15)

A storage unit that stores a plurality of pieces of failure information; and an arithmetic processing unit that creates a failure tree.
The arithmetic processing unit includes:
An information refinement unit that extracts failure information to be used for creating a failure tree from the plurality of failure information stored in the storage unit based on a user instruction;
A fault tree creating unit that creates a fault tree from the fault information extracted by the information narrowing unit;
Fault tree creation device including Of the fault trees created by the fault tree creation unit, a fault tree selection unit that allows the user to select a fault tree to be stored;
A fault tree storage unit that stores the fault tree selected by the user in the storage unit;
The fault tree creation device according to claim 1, further comprising: The information narrowing unit receives input of search conditions from the user, and extracts failure information that matches the search conditions from the plurality of pieces of failure information.
The fault tree creation device according to claim 2. The fault tree storage unit stores the fault tree in the storage unit, and stores the search condition used when the information narrowing unit extracts the fault information from which the fault tree was created. To save,
The fault tree creation apparatus according to claim 3. The information narrowing unit reads the search condition stored in the storage unit based on the user's instruction, and extracts failure information that matches the read search condition from the plurality of failure information To
The fault tree creation apparatus according to claim 4. The failure information includes information on the content of the failure of the railway vehicle, and information on the weather and the travel section when the failure occurs in the railway vehicle,
The search condition is a condition related to the weather and a traveling section.
The fault tree creation apparatus according to claim 3. The fault tree creation device receives a search condition held by the other fault tree creation device from another fault tree creation device;
Of the received search conditions, the search conditions that are not stored in the storage unit are stored in the storage unit.
The fault tree creation apparatus according to claim 3, wherein: The fault tree creation device receives a fault tree held by the other fault tree creation device from another fault tree creation device;
Of the received fault trees, the fault tree that is not stored in the storage unit is stored in the storage unit.
The fault tree creation device according to claim 2, wherein: The storage unit is shared with other fault tree creation devices,
The fault tree creation apparatus according to claim 1, wherein: A computer having a storage unit that stores a plurality of pieces of failure information and an arithmetic processing unit that creates a failure tree,
Based on a user instruction, an information narrowing step of extracting failure information to be used for creating a failure tree from the plurality of failure information stored in the storage unit;
A fault tree creating step of creating a fault tree from the fault information extracted by the information narrowing unit;
The fault tree creation method is executed. Of the fault trees created in the fault tree creation step, a fault tree selection step that allows the user to select a fault tree to be stored;
A fault tree storing step of storing the fault tree selected by the user in the storage unit;
The fault tree creation method according to claim 10, further comprising: In the information narrowing-down step, an input of a search condition is received from the user, and failure information that matches the search condition is extracted from the plurality of failure information.
The fault tree creation method according to claim 11. In the fault tree storing step, the search condition used when the fault narrowing section is stored in the storage unit and the information narrowing unit extracts the fault information that is the basis of the fault tree creation is stored in the storage unit. To save,
The fault tree creation method according to claim 12. In the information narrowing-down step, the search condition stored in the storage unit is read based on an instruction from the user, and failure information that matches the read search condition is selected from the plurality of failure information. Extract,
The fault tree creation method according to claim 13. The failure information includes information on the content of the failure of the railway vehicle, and information on the weather and the travel section when the failure occurs in the railway vehicle,
The search condition is a condition related to the weather and a traveling section.
The fault tree creation method according to claim 12.

Images