Maintenance operation support system and maintenance operation method

Abstract

To provide a maintenance support system that increases the operating efficiency of a maintenance business operator coping with the occurrence or onset of an abnormality in industrial machinery and industrial facilities, and can quickly return the industrial machinery and industrial facilities to a normal state.SOLUTION: A maintenance operation support system 1 comprises: a fundamental event inference engine 20 that infers a fundamental event 42 to be the cause of a trouble in industrial machinery 2 and the probability of occurrence of the event according to a causal model created in advance; a checking operations presentation engine 21 that presents checking operations for checking whether the fundamental event 42 occurs according to a processing model created in advance; a history information extraction engine 22 that extracts, from a history management database 31, the history of environmental related information related to the operating environment of the industrial machinery 2 for which the checking operations are carried out and the history of maintenance operations records; and a fundamental event specification engine 23 that specifies the fundamental event 42 in which the causal relationship occurs with a top event 40 or an intermediate event 41 according to the processing model created in advance.SELECTED DRAWING: Figure 3

Description

[Technical Field]

[0001] The technology disclosed in this specification relates to a maintenance work support system and a maintenance work operation method. [Background technology]

[0002] Patent Document 1 discloses a technique for remotely managing a device to be managed based on the status of the device to be managed detected by a detection unit. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2021-162999 Summary of the Invention [Problem to be solved by the invention]

[0004] When an abnormality occurs or a sign of an abnormality is observed in industrial machinery and facilities, maintenance is carried out by a maintenance company. In order to prevent a decline in the availability of industrial machinery and facilities, it is necessary to quickly restore them to normal operation. There is also a need to improve the operational efficiency of maintenance companies.

[0005] The technology disclosed in this specification provides a maintenance support system that can improve the work efficiency of maintenance companies that respond to the occurrence and signs of abnormalities in industrial machinery and industrial equipment, etc., while quickly restoring the equipment to a normal state. [Means for solving the problem]

[0006] This specification discloses a maintenance work support system. The maintenance work support system includes: a basic event inference engine that, by inputting failure event information related to a top event or intermediate event related to a failure of an industrial machine, infers one or more basic events that cause the failure and their occurrence probabilities according to a pre-created causal model; a confirmation work presentation engine that, by inputting the basic events and their occurrence probabilities inferred by the basic event inference engine, presents confirmation works for confirming whether the basic events have occurred in order of the basic event's occurrence probability, according to a pre-created processing model; a history information extraction engine that, by inputting identification information of the industrial machine in which the failure has occurred, extracts from a history management database history information that can be used for the confirmation work, the history of environment-related information related to the operating environment of the target industrial machine and the history of maintenance work records; and a basic event identification engine that, by inputting the results of the required confirmation works presented by the confirmation work presentation engine, identifies basic events that have a causal relationship with the top event or intermediate event, according to the pre-created processing model. [Effects of the Invention]

[0007] The technology disclosed in this specification makes it possible to provide a maintenance support system that can improve the work efficiency of maintenance companies that respond to the occurrence and signs of abnormalities in industrial machinery and industrial equipment, while quickly restoring the equipment to a normal state. [Brief explanation of the drawings]

[0008] [Figure 1] FIG. 1 is a diagram schematically illustrating a maintenance work support system according to an embodiment. [Figure 2] FIG. 2 is a hardware configuration diagram showing the information processing apparatus according to the embodiment. [Figure 3] FIG. 3 is a functional block diagram showing a maintenance work support system according to the embodiment. [Figure 4] FIG. 4 is a diagram for explaining the fault tree analysis according to the embodiment. [Figure 5]FIG. 5 is a diagram for explaining a resource inference model according to the embodiment. [Figure 6] FIG. 6 is a flowchart showing a maintenance work support method according to the embodiment. [Figure 7] FIG. 7 is a diagram illustrating an example of a maintenance work support portal screen according to the embodiment. [Figure 8] FIG. 8 is a diagram for explaining a maintenance work operation method according to the embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0009] [Maintenance Work Support System Overview] FIG. 1 is a diagram schematically illustrating a maintenance work support system 1 according to an embodiment. The maintenance work support system 1 supports maintenance work performed by a maintenance business operator. An industrial machine 2 is installed at a business establishment 3. The maintenance business operator provides various maintenance services to the industrial machine 2. The maintenance services include after-maintenance, which is performed after an abnormality occurs in the industrial machine 2, and before-maintenance, which is performed before an abnormality occurs in the industrial machine 2. Before-maintenance includes time-based maintenance and condition-based maintenance. Condition-based maintenance is considered the most preferable maintenance method because it reduces unnecessary work compared to time-based maintenance. The maintenance work support system 1 mainly supports work related to after-maintenance, which is highly urgent and tends to require a large amount of work effort, thereby reducing the burden on the operators (service personnel).

[0010] Industrial machinery 2 refers to machinery and equipment used to produce goods or provide services. Industrial machinery 2 mainly includes utility conversion machinery and equipment that converts primary utilities into secondary utilities that can be used by demand facilities.

[0011] Utilities refer to energy sources or fluids required for industrial activity. Examples of primary utilities input to utility conversion machines and equipment include fuel (gas, oil), electricity, and raw water. Examples of secondary utilities output from industrial machines 2 include heat transfer media (steam, thermal oil, hot water, cold water), compressed air, electricity, and treated water.

[0012] Demand facilities use the secondary utilities output from utility conversion machinery and equipment. For example, demand facilities use heat transfer media as a heat source for various production processes or air conditioning. Demand facilities use compressed air to power pneumatic equipment or pneumatic tools. Demand facilities use electricity to power electrically powered equipment, power tools, or lighting. Demand facilities use treated water as process water for food, cosmetics, pharmaceuticals, or semiconductor manufacturing.

[0013] Examples of utility conversion machinery and equipment include thermal equipment, air compressors, generators, and water treatment equipment. Examples of thermal equipment include combustion-type steam boilers, electric heater steam boilers, heat recovery steam boilers, combustion-type heat medium boilers, combustion-type hot water boilers, electric heater hot water boilers, heat recovery hot water boilers, electric heat pumps, electric chillers, electric heat pump chillers, and flash steam generators. Examples of air compressors include electric air compressors, heat recovery electric air compressors, and steam-driven air compressors. Examples of generators include mono-generation generators and co-generation generators. Examples of water treatment equipment include reverse osmosis membrane devices, water softeners, deoxygenators, and various filtration devices.

[0014] In addition, industrial machinery 2 may include medical machinery and equipment used in a series of processes from receiving items to be cleaned and sterilized to unloading them, laundry machinery and equipment used in a series of processes from collecting items to shipping them, food and beverage manufacturing machinery and equipment used in a series of processes from receiving raw materials to storing products, vehicles that transport cargo unmanned between multiple points set up within a business premises, navigation machinery and equipment installed on ships such as cargo ships, etc.

[0015] Examples of medical devices include washer and sterilizer. Examples of washer include vacuum boiling washer and ultrasonic washer. Examples of sterilizer include steam sterilizer and gas sterilizer.

[0016] Examples of laundry appliances include washing machines, dryers, and finishing machines. Examples of washing machines include continuous washing machines, cold water washing machines, and dry cleaning machines. Examples of dryers include gas dryers and steam dryers. Examples of finishing machines include gas roll ironers and steam roll ironers.

[0017] Examples of machinery and equipment for producing food and beverages include thawing machines, cooking machines, cooling machines, and sterilizers. Examples of thawing machines include vacuum steam thawing machines, microwave thawing machines, high-frequency thawing machines, and running water thawing machines. Examples of cooking machines include steam kneaders, steam kettles, and saturated steam cookers. Examples of cooling machines include vacuum coolers, cold water coolers, and cold air coolers. Examples of sterilizers include retort sterilizers and pasteurizers.

[0018] Examples of autonomously travelling unmanned guided vehicles include a trolley vehicle, a forklift vehicle and a tow vehicle.

[0019] Examples of navigational machinery and equipment include main engines (single-fuel / dual-fuel diesel engines), turbochargers (auxiliary machinery), exhaust gas economizers (steam generators), shaft generators, steam turbine generators, binary generators, freshwater boilers, BOG combustion boilers, and exhaust gas scrubbing equipment. Examples of cargo handling machinery and equipment include diesel generators, cranes, derricks, ballast water pumps, and ballast water treatment systems.

[0020] An establishment 3 refers to an individual location where the production of goods or the provision of services is carried out as a business. Industrial machinery 2 is installed at establishment 3. If a factory that produces goods, etc. is set up at establishment 3, industrial machinery 2 is installed in the factory. Examples of factories include food factories, beverage factories, metal product factories, plastic product factories, textile factories, and laundry factories.

[0021] It should be noted that the establishment 3 that provides the service does not necessarily have to have a factory. The business conducted at the establishment 3 may include public health services. Examples of public health services include hospitals, clinics, and health centers. The establishment 3 may also include a food service center. Furthermore, instead of the establishment 3, the industrial machine 2 may be installed on a ship used for maritime transportation services.

[0022] The industrial machine 2 is provided with an environmental sensor 4. The environmental sensor 4 detects environmental information of the business establishment 3 in which the industrial machine 2 is installed. Typical environmental information is the environmental state or environmental conditions of the space in which the industrial machine 2 operates (environment-related information related to the operating environment). The environmental information includes physical parameters of the main body of the industrial machine 2 and its surroundings. Data detected by some of the environmental sensors 4 is used for the operation or control of the industrial machine 2. Examples of environmental sensors 4 include a temperature sensor, a humidity sensor, a pressure sensor, a water level sensor, a flow rate sensor, an electrical conductivity sensor (EC sensor), a power sensor, a distance sensor, an image sensor, and a force sensor.

[0023] The maintenance work support system 1 includes a server 5 and an information processing device 6. The server 5 has a database related to maintenance. The server 5 receives environmental information detected by an environmental sensor 4 via a communication network, which will be described later.

[0024] An environmental information collection system for efficiently collecting vast amounts of environmental information may be installed in the business establishment 3. The environmental information collection system is preferably configured to acquire and store environmental information detected by, for example, multiple environmental sensors 4, and includes multiple information processing devices (industrial controllers, edge computers, etc.) with a hierarchical structure that transmits environmental information from downstream to upstream. An industrial controller is a device that has the function of controlling the operation of industrial machinery 2 and controlling data collection, and examples of such devices include microcomputers (mCUs) and programmable logic controllers (PLCs). If the industrial controller has a function specialized for data collection, it can also be configured as a data collection terminal to which multiple environmental sensors 4 can be connected.

[0025] In the environmental information collection system, the server 5 is placed at the top of a plurality of hierarchically structured information processing devices. The server 5 has an information accumulation platform. The information accumulation platform is, for example, an open IoT operating system based on cloud computing, and is capable of systematically accumulating environmental information relating to the operating environment of the industrial machinery 2 and maintenance work records. Note that some or all of the functions realized by the information accumulation platform may be incorporated into an edge computer installed in the business establishment 3.

[0026] The information processing device 6 performs information processing to support maintenance work. The information processing device 6 manages the worker 60 (serviceman) who performs the maintenance. The information processing device 6 provides useful information and reference information related to maintenance to the information terminal 7 carried by the worker 60 via a communication line. Examples of the information terminal 7 include a smartphone and a tablet terminal. The information terminal 7 may also be a personal computer. The information terminal 7 may also be able to access the server 5 via the communication line.

[0027] The information processing device 6 manages items 61 used for maintenance. The items 61 include parts and tools used for maintenance of the industrial machine 2. The items 61 also include a warehouse where the parts and tools are stored.

[0028] The information processing device 6 manages content 62 used for maintenance. Examples of the content 62 include written manuals (procedure manuals, instructions, etc.), video manuals, and past maintenance reports.

[0029] An input device 8 and a display device 9 are connected to the information processing device 6. The input device 8 generates input data when operated by a user of the maintenance work support system 1. The input data generated by the input device 8 is input to the information processing device 6. Examples of the input device 8 include a touch panel (touch sensor), a computer keyboard, and a voice input device. Note that the input data may also be input to the information processing device 6 via a communication network. The display device 9 provides display data to the user of the maintenance work support system 1. The display device 9 displays the display data transmitted from the information processing device 6. Examples of the display device 9 include a flat panel display such as a liquid crystal display or an organic EL display.

[0030] [computer] 2 is a hardware configuration diagram showing an information processing device 6 according to an embodiment. The information processing device 6 includes a computer 10. The computer 10 has a processor 11, a storage device 12, a communication interface 13, and an input / output interface 14. The information processing device 6 also has a power supply device (not shown). Like the information processing device 6, the server 5 also includes a computer 10.

[0031] The processor 11 includes a CPU (Central Processing Unit). The processor 11 may also include a GPU (Graphics Processing Unit). The storage device 12 includes a recording medium that stores computer programs and data in a manner that allows the processor 11 to read them. The storage device 12 includes on-board system memory such as RAM (Random Access Memory) or ROM (Read Only Memory), large-capacity flash memory such as an SD card or USB memory, and large-capacity storage such as an HDD (Hard Disk Drive) or SSD (Solid State Drive).

[0032] The communication interface 13 communicates via a communication network. Examples of communication networks include a local area network (LAN), a wide area network (WAN), and a commercial network such as the Internet. The local area network may be a wired LAN or a wireless LAN. The wide area network may include a mobile line or a satellite communication line. The computer 10 transmits data to an external computer via the communication network. The computer 10 receives data from an external computer via the communication network. The computer 10 is connected to an external device via the input / output interface 14.

[0033] The storage device 12 stores application software 15. The application software 15 is an example of a computer program. The processor 11 reads the application software 15 from the storage device 12, loads it into a system memory, and executes processing in accordance with the application software 15. The application software 15 may be distributed to the computer 10 via a communication network.

[0034] [Maintenance work support system functions] 3 is a functional block diagram showing the maintenance work support system 1 according to the embodiment. The maintenance work support system 1 has a plurality of engines, a plurality of applications, and a plurality of databases.

[0035] An engine is a processing mechanism that provides a function for performing specific information processing using a computer. The engine may be a computer program or hardware. In the embodiment, the engine is provided by the processor 11 of the information processing device 6.

[0036] An application refers to a processing function provided by application software 15 installed on computer 10 .

[0037] A database is a collection of information that is collected and organized on a computer.

[0038] The maintenance work support system 1 includes a basic event inference engine 20, a confirmation work presentation engine 21, a history information extraction engine 22, a basic event identification engine 23, a resource inference engine 24, a knowledge information extraction engine 25, an item procurement application 26, a personnel selection application 27, a content extraction application 28, an action support application 29, a model update application 30, a history management database 31, an item management database 32, a personnel allocation management database 33, a content management database 34, and a report management database 35.

[0039] The basic event inference engine 20 inputs failure event information related to a top event 40 or intermediate event 41 related to a failure of the industrial machine 2, and infers one or more basic events 42 that cause the failure and their occurrence probability according to a causal model created in advance. Causal models include a Bayesian network (described below) and a Bayesian tree. A Bayesian network is a graphical model that describes complex causal relationships using probability.

[0040] The malfunction event information of the industrial machine 2 is a specified alarm code issued by the industrial machine 2 or the details of the malfunction event expressed in words. Note that the malfunction event information of the industrial machine 2 may be detection data of the environmental sensor 4.

[0041] FIG. 4 is a diagram for explaining fault tree analysis according to an embodiment. Fault tree analysis (FTA) is an analytical technique that identifies an undesirable event as a top event 40 and identifies all routes leading to the top event 40. The routes are shown in a logical tree diagram and are also called "fault tree analysis." Fault tree analysis is widely used to analyze the causes of various malfunctions (such as abnormalities, failures, and defects) and to calculate the probability of their occurrence.

[0042] In fault tree analysis, an intermediate event 41 (deployment event) is an event that occurs as a combination of basic events 42 and then develops further. A basic event 42 is a basic event that does not develop any further. A non-deployment event 43 is an event that cannot be developed any further due to various reasons such as a lack of information or a lack of worker ability. An AND gate 44 is logic that causes a higher-level event to occur when all lower-level events occur simultaneously. An OR gate 45 is logic that causes a higher-level event to occur when any of the lower-level events occur.

[0043] A causal model is an inference model that combines a rule (knowledge)-based model in which the mechanisms of defects are known academically or empirically, and a machine learning model created by machine learning (supervised learning) that uses past maintenance reports (documents containing features such as the defect event, cause of occurrence, treatment details, and whether the defect was resolved or not) as training data. The cause of occurrence (basic event) derived by the causal model can be single or multiple. The latter is usually the case, and an occurrence probability is attached. The causal model is updated from time to time as new maintenance reports are learned, and the occurrence probability of a particular defect event will change.

[0044] The confirmation task presentation engine 21 inputs the basic events 42 inferred by the basic event inference engine 20 and their occurrence probabilities, and presents confirmation tasks for confirming whether or not the basic events have occurred in order of highest occurrence probability, according to a processing model created in advance.

[0045] A pre-created flowchart is an example of a processing model, but the processing model can be created arbitrarily.

[0046] The historical information extraction engine 22 inputs the identification information of the industrial machine 2 in which a malfunction has occurred, and extracts from the history management database 31 historical information that can be used for confirmation work, such as the history of environmental information related to the operating environment of the target industrial machine 2 and the history of maintenance work records.

[0047] The identification information (ID) of the industrial machine 2 is a unique machine number (machine number) assigned to each individual industrial machine 2 or the like.

[0048] The history of environment-related information is, for example, information accumulated over a certain period of time regarding the state and conditions (temperature, pressure, flow rate, water quality, etc.) of the operating environment of the industrial machine 2. The environment-related information is mainly detected in chronological order by the environmental sensor 4, and may include characteristic values ​​and performance values ​​calculated based on multiple detected values.

[0049] The history of maintenance work records is information that is continuously accumulated until the industrial machinery 2 is disposed of, such as a trial run report created during the trial run after the introduction of the industrial machinery 2, an inspection report created during past regular inspections, and a maintenance report created when dealing with past abnormality signs or when dealing with abnormalities that have occurred.

[0050] The basic event identification engine 23 inputs the results of the required confirmation work presented by the confirmation work presentation engine 21, and identifies basic events 42 that have a causal relationship with the top event 40 or intermediate event 41 according to a processing model created in advance.

[0051] An example of a process model is a process map such as a pre-created flowchart, etc. The process model can be created in any style other than a process map.

[0052] The resource inference engine 24 infers the resources required for the work to resolve the basic event 42 identified by the basic event identification engine 23 in accordance with a resource inference model 50 created in advance.

[0053] 5 is a diagram illustrating a resource inference model 50 according to an embodiment. As shown in FIG. 5, the resource inference model 50 includes at least an item inference model 51 that infers items 61 required for resolving the basic events 42, a human resources inference model 52 that infers the capabilities of workers 60 required for resolving the basic events 42, and a content inference model 53 that infers content 62 useful for resolving the basic events 42.

[0054] Work to resolve basic event 42 includes replacing parts, restoring parts (cleaning dust filters and strainers, cleaning heat exchangers and separation membranes, etc.), and replacing or replenishing consumables (replenishing lubricating oil, replenishing water treatment chemicals, replacing reagent cartridges, replacing pretreatment filters, replacing water treatment materials, etc.).

[0055] The item procurement application 26 checks the item 61 inferred by executing the item inference model 51 against the item management database 32 and procures the item 61.

[0056] The item management database 32 comprehensively manages parts and tools stocked in warehouses (headquarters logistics center, main area logistics center, maintenance base, etc.) owned by the maintenance company in charge of maintenance work, warehouses owned by maintenance business partners that complement maintenance work, warehouses owned by main parts manufacturers, warehouses owned by alternative parts manufacturers, etc. Procuring an item means securing the item if it is in stock with the maintenance company or maintenance business partner, and if it is not in stock, ordering it from the manufacturer or receiving it in stock.

[0057] The personnel selection application 27 compares the capabilities of the workers 60 inferred by executing the personnel inference model 52 with the personnel allocation management database 33, and selects the workers 60 to be assigned to the resolution work.

[0058] The personnel allocation management database 33 mainly manages the bases to which workers 60 (field engineers who can be dispatched by the maintenance company in charge of maintenance work and maintenance business partners who complement the maintenance work) who are seconded to the site belong, as well as the ability information of the workers 60 (qualifications (public exams, company exams, etc.), work experience, and proficiency). The workers 60 also include personnel who remotely support on-site work.

[0059] The content extraction application 28 compares the content 62 inferred by executing the content inference model 53 with the content management database 34, and extracts the content 62 to be referenced when performing resolution work.

[0060] The content 62 managed by the content management database 34 includes written manuals (procedure manuals, instructions), video manuals, past maintenance reports, and the like.

[0061] The action support application 29 incorporates the preparation information on the items 61 procured by the item procurement application 26 and the reference information on the content 62 extracted by the content extraction application 28 into the daily action plan information of the workers 60 selected by the personnel selection application 27. The daily action plan information for each worker 60 is stored, for example, on the server 5 in an updated state, and can be referenced via the information terminal 7 carried by the worker 60.

[0062] The report management database 35 accumulates work reports (maintenance reports) that record the details of the resolution work performed by the workers 60 selected by the personnel selection application 27. The workers 60 operate, for example, the input device of the information terminal 7 to input the details of the resolution work into the information terminal 7. The work reports that record the details of the resolution work input into the information terminal 7 are accumulated in the report management database 35 via the information processing device 6.

[0063] The knowledge information extraction engine 25 extracts knowledge information useful for solving the basic event 42 from the work reports stored in the report management database 35 .

[0064] The knowledge information extraction engine 25 selects cases in which the basic event 42 was successfully resolved from the work report, and using a pre-created language model, reads characteristic text contained in the records of the successful cases, performs natural language processing, and outputs the generated sentences as knowledge information.

[0065] The knowledge information extraction engine 25 selects cases in which defects have been successfully resolved from among the work reports newly registered in the report management database 35, and performs feature engineering to extract the knowledge information contained in the records of the successful cases, such as the top event 40 that occurred (defect content), the details of the measures taken by the worker (including additional confirmation work based on historical information, adjustment work not included in the content, and other items that could become new know-how), and the finally determined basic event 42 (cause of the defect).

[0066] A causal model that has already been created cannot make sufficient inferences about basic events 42 for which knowledge has not been gained from past cases. By updating the causal model through machine learning using newly acquired knowledge information as training data, it is possible to cover basic events 42 and improve the accuracy of occurrence probability.

[0067] In addition, the knowledge information extraction engine 25 utilizes the newly obtained knowledge information to update the processing model in the confirmation task presentation engine 21, the processing model in the basic event identification engine 23, and the inference model in the resource inference engine 24, thereby improving work efficiency.

[0068] The model update application 30 inputs knowledge information extracted by the knowledge information extraction engine 25 to update at least one of the causal model in the basic event inference engine 20, the processing model in the confirmation task presentation engine 21, the processing model in the basic event identification engine 23, and the resource inference model 50 in the resource inference engine 24.

[0069] [Maintenance work support method] Fig. 6 is a flowchart showing a maintenance work support method according to an embodiment. Fig. 7 is a diagram showing an example of a maintenance work support portal screen (hereinafter simply referred to as "portal screen 90") displayed on the display device 9 or the information terminal 7. The portal screen 90 functions as a user interface (UI) for the worker 60 (serviceman, etc.). As an example, a method for supporting after-maintenance of a reverse osmosis membrane device (RO membrane device) used for pure water production, etc. will be described below.

[0070] A portal screen 90 generated by the information processing device 6 displays a first text box 91 for entering information about a malfunction that has occurred in the RO membrane device. Malfunction event information related to a top event 40 or intermediate event 41 related to the malfunction is entered in the first text box 91 via the input device 8 or the information terminal 7. Specifically, either an alarm code or the content of the malfunction (e.g., a sentence) is entered in the first text box 91. After entering the malfunction event information, pressing an execute button 91a located near the first text box 91 inputs the information in the first text box 91 into the basic event inference engine 20. Based on the input malfunction event information, the basic event inference engine 20 infers a basic event 42 that is the cause of the malfunction and its occurrence probability in accordance with a causal model (step S1).

[0071] In the fault tree analysis of an RO membrane device, one of the critical peak events 40 is an abnormality in the water quality of the treated water. When an abnormality in the water quality of the treated water occurs, the RO membrane device is shut down to prevent adverse effects on the production process of the demand facility. Intermediate events 41 that cause the peak event 40 include false detection (the actual water quality is normal), overconcentration (a decrease in the solute removal rate), membrane deterioration (loosening of the skin layer), and membrane damage (peeling of the skin layer).

[0072] Basic events 42 that can lead to intermediate events 41 include EC sensor failure for false detection, drain valve failure (insufficient drainage flow rate), temperature sensor failure (improper adjustment of recovery rate), and deterioration of raw water quality (improper adjustment of recovery rate) for overconcentration, residual chlorine leak (breakthrough of activated carbon) for membrane deterioration, and back pressure for membrane damage. Non-deployment events 43 include events where the cause is difficult to determine, such as material contamination.

[0073] A list of basic events 42 inferred by the basic event inference engine 20 is displayed in a first window 92 of the portal screen 90. In the first window 92, the basic events 42 are displayed in descending order of occurrence probability. In the example shown in FIG. 7, the first window 92 displays the aforementioned "EC sensor failure," "drain valve failure," "deterioration of raw water quality," "residual chlorine leak," "occurrence of back pressure," and the like as basic events 42 to be inferred. The first window 92 also displays the numerical value of the occurrence probability for each basic event 42.

[0074] The basic events 42 and their occurrence probabilities inferred by the basic event inference engine 20 in step S1 are input to the confirmation task presentation engine 21. Based on the basic events 42 and their occurrence probabilities inferred by the basic event inference engine 20, the confirmation task presentation engine 21 presents confirmation tasks for confirming whether or not the basic events 42 have occurred in order of the highest occurrence probability of the basic events, in accordance with a processing model created in advance (step S2).

[0075] A list of confirmation tasks presented by the confirmation task presentation engine 21 is displayed in the second window 93 of the portal screen 90. In the second window 93, the confirmation tasks are displayed in order of their corresponding basic events 42 with a high probability of occurrence. In the example shown in FIG. 7, the second window 93 displays "Check the purity of the treated water with a handy EC meter" and "Check the EC display value of the device" as confirmation tasks for "EC sensor failure." The second window 93 also displays "Check whether the drain valve is disconnected or broken" and "Check whether the drain valve is clogged" as confirmation tasks for "drain valve failure."

[0076] An example of a check for a "temperature sensor failure" is "checking the abnormality detection history of the raw water temperature sensor." An example of a check for a "deterioration in raw water quality" is "checking whether the silica meter has run out of reagent." Examples of checks for a "residual chlorine leak" are "checking the history of residual chlorine meter concentration abnormality alarms," ​​"checking the history of residual chlorine meter measurements," and "checking whether the residual chlorine meter has run out of reagent." Examples of checks for "back pressure generation" are "checking the installation status of the secondary piping (presence or absence of riser piping, bent piping, junction piping, etc.)" and "checking the installation status of the primary piping (presence or absence of faller piping, etc.)."

[0077] The portal screen 90 displays a third window 94 showing the order of inspection work in the RO membrane equipment in a flowchart, and a fourth window 95 showing the areas subject to the inspection work in drawings and photographs. These flowcharts, drawings, photographs, etc. are presented by the inspection work presentation engine 21 together with a list of inspection work. The worker 60 can smoothly carry out the inspection work by referring to the information in the third window 94 and the fourth window 95.

[0078] The portal screen 90 also displays a second text box 96 for entering identification information of the RO membrane device in which the malfunction event occurred. Specifically, the machine number is entered in the second text box 96 via the input device 8 or the information terminal 7. After entering the identification information, when a search button 96a arranged near the second text box 96 is pressed, the information in the second text box 96 is input into the history information extraction engine 22. Based on the input identification information, the history information extraction engine 22 extracts from the history management database 31 history information that can be used for confirmation work, such as a history of environment-related information related to the operating environment of the target industrial machine 2 and a history of maintenance work records (step S3).

[0079] The historical information extracted by the historical information extraction engine 22 can be called up using multiple call buttons 97a arranged on the portal screen 90. The environmental information history includes a system flow diagram of the RO membrane device (an equipment flow diagram including construction information), water analysis results of raw water, etc., as well as monitor data of the RO membrane device itself and monitor data of a water quality meter attached to the pretreatment device. The monitor data is, for example, environmental information accumulated in a storage device of an industrial controller or server 5. The monitor data may also be environmental information acquired in real time from the environmental sensor 4. The maintenance work record history includes past reports (maintenance reports, periodic inspection reports). Pressing a required call button 97a generates a link in the fifth window 97 for viewing the corresponding historical information.

[0080] System flow diagrams, which are a history of environmental information, are used as a useful set of information for efficiently carrying out verification work. For example, by reading the construction information for the RO membrane equipment and surrounding areas from the system flow diagram, it may be possible to check whether piping construction that induces back pressure has been carried out without having to visit the site.

[0081] Water analysis results and monitoring data, which are environmental history records, are useful for efficiently conducting verification work. The first example is the history of residual chlorine concentration. Residual chlorine leaks due to breakthrough of the activated carbon filter installed upstream of the RO membrane unit can lead to membrane degradation. Both sudden alarm-level leaks and continuous, minute leaks that do not trigger an alarm can be cause for concern. The second example is the history of permeation flux. When the skin layer loosens due to membrane degradation, the water permeation flux increases, resulting in increased solute leakage. The third example is the history of raw water quality analysis results. When the electrical conductivity of raw water deteriorates and increases during RO membrane operation, overconcentration occurs on the membrane surface, increasing solute leakage. The fourth example is the status history of a residual chlorine concentration meter. If the residual chlorine concentration meter runs out of reagent, it will no longer be able to measure, making it impossible to detect leaks. The fifth example is the status history of a silica concentration meter. If the silica concentration meter runs out of reagent, it will no longer be able to measure, making it impossible to adjust the recovery rate (i.e., concentration ratio). A sixth example is the abnormality detection history of the raw water temperature sensor. If the temperature sensor malfunctions, the recovery rate cannot be adjusted to the appropriate value. By referring to the water analysis results and monitor data in this way, it is possible to check for residual chlorine leaks and whether the raw water quality is deteriorating without visiting the site.

[0082] Past reports, which are maintenance work records, are used as a useful set of information to efficiently carry out confirmation work. For example, by referring to the most recent regular inspection report, it is possible to determine whether the functional parts of the RO membrane equipment (drain valves, membrane elements, etc.) are operating normally. Also, by referring to the most recent maintenance report, it is possible to determine whether a malfunctioning functional part (drain valve, membrane element, etc.) has been repaired or replaced. In this way, by referring to past reports, it is sometimes possible to determine whether the condition of functional parts is good or bad without having to visit the site.

[0083] The worker 60 or support personnel performs the required confirmation work presented by the confirmation work presentation engine 21 in step S2 while referring to the history of the environment-related information and the history of the maintenance work records. The results of the confirmation work can be entered into a result input box 98 on the portal screen 90 via the input device 8 or the information terminal 7. The results can be entered either by entering a numerical value or by entering a pass / fail judgment value. After entering the results, pressing a judgment button 98a located near the result input box 98 inputs the information in the result input box 98 into the basic event identification engine 23. Based on the input results of the confirmation work, the basic event identification engine 23 identifies basic events 42 that have a causal relationship with the top event 40 or intermediate event 41 in accordance with a pre-created processing model (step S4). The identified basic events 42 can be easily recognized by the worker 60 in the second window 93 by displaying a mark or identifier, highlighting the text or background color, or other methods.

[0084] The portal screen 90 displays a sixth window 99 for viewing related materials that are useful when working to resolve the identified basic event 42. The related materials are presented by the basic event identification engine 23, and include various manuals, collections of trouble cases, maintenance reports for similar malfunction event information, etc. Links for calling up these related materials are generated in the sixth window 99.

[0085] The resource inference engine 24 infers resources required for the resolution of the basic event 42 identified by the basic event identification engine 23 in step S4, according to a resource inference model 50 created in advance. The resource inference engine 24 infers, as resources, items 61 required for the resolution of the basic event 42, the capabilities of workers 60 required for the resolution of the basic event 42, and content 62 useful for the resolution of the basic event 42 (step S5).

[0086] The item procurement application 26 compares the items 61 inferred by the execution of the item inference model 51 with the item management database 32 and procures the items 61. The personnel selection application 27 compares the capabilities of the workers 60 inferred by the execution of the personnel inference model 52 with the personnel allocation management database 33 and selects the workers 60 to be involved in the resolution work. The content extraction application 28 compares the content 62 inferred by the execution of the content inference model 53 with the content management database 34 and extracts the content 62 to be referenced when the resolution work is carried out (step S6).

[0087] The behavioral support application 29 incorporates the preparation information regarding the items 61 procured by the item procurement application 26 and the reference information regarding the content 62 extracted by the content extraction application 28 into the daily behavioral plan information of the worker 60 selected by the personnel selection application 27 (step S7).

[0088] That is, when the maintenance schedule for the selected worker 60 is displayed on the information terminal 7 carried by the selected worker 60, the items 61 to be brought on the day of maintenance and the content 62 that is likely to be referenced on that day are automatically registered in the schedule. By checking the schedule, the selected worker 60 can confirm the items 61 to be brought on the day of maintenance and the content 62 that can be used.

[0089] The worker 60 carries out maintenance while carrying the information terminal 7. When performing work to resolve the identified basic event 42, the worker 60 performs the work while referring to related materials in the display data 97 on the portal screen and the content 62 incorporated into the daily action plan information. Examples of the content of the resolution work include "replacement of the EC sensor," "replacement of the drain valve, replacement of the raw water temperature sensor, replacement of the silica meter reagent," "replacement of the membrane element, replacement of the residual chlorine meter reagent, replacement of the activated carbon," and "redoing the piping installation."

[0090] The worker 60 operates, for example, the input device of the information terminal 7 to input the details of the resolution work into the information terminal 7. A work report recording the details of the resolution work input into the information terminal 7 is registered in the report management database 35 via the information processing device 6. The report management database 35 accumulates a plurality of work reports (step S8).

[0091] The knowledge information extraction engine 25 extracts knowledge information useful for resolving the basic event 42 from the work reports stored in the report management database 35. The knowledge information extraction engine 25 extracts knowledge information useful for resolving the basic event 42 from the work reports stored in the report management database 35 (step S9).

[0092] The knowledge information extraction engine 25 may select cases in which the basic event 42 was successfully resolved from the work reports, and use a pre-created language model to read characteristic text contained in the records of the successful cases, perform natural language processing, and output the generated sentences as knowledge information.

[0093] In step S9, the knowledge information extracted by the knowledge information extraction engine 25 is input to the model update application 30. Based on the knowledge information, the model update application 30 updates at least one of the causal model in the basic event inference engine 20, the processing model in the confirmation task presentation engine 21, the processing model in the basic event identification engine 23, and the resource inference model 50 in the resource inference engine 24 (step S10).

[0094] [Maintenance operation method] 8 is a diagram for explaining a maintenance work operation method according to the embodiment. The information processing device 6 utilizes the maintenance work support system 1 to implement a PDCA (Plan, Do, Check, Act) cycle.

[0095] As shown in FIG. 8, the planning process is a process involving the identification of basic events using a basic event inference engine 20, a confirmation task presentation engine 21, and a basic event identification engine 23.

[0096] The execution process is the process that involves performing the resolution task using the resource inference engine 24 and the behavior support application 29 .

[0097] The evaluation process involves the extraction of useful knowledge information using the knowledge information extraction engine 25 .

[0098] The refinement process involves updating the model using a model update application 30 .

[0099] [effect] As described above, in the embodiment, the maintenance work support system 1 includes: a basic event inference engine 20 that, by inputting malfunction event information related to a top event 40 or an intermediate event 41 related to a malfunction of an industrial machine 2, infers one or more basic events 42 that cause the malfunction and their occurrence probability in accordance with a pre-created causal model; a confirmation work presentation engine 21 that, by inputting the basic event 42 inferred by the basic event inference engine 20 and its occurrence probability, presents confirmation works for confirming whether the basic event 42 has occurred in order of highest occurrence probability, in accordance with a pre-created processing model; a history information extraction engine 22 that, by inputting identification information of the industrial machine 2 in which a malfunction has occurred, extracts from the history management database 31 history information that can be used for the confirmation work, the history of environment-related information related to the operating environment of the target industrial machine 2 and the history of maintenance work records; and a basic event identification engine 23 that, by inputting the results of the required confirmation work presented by the confirmation work presentation engine 21, identifies a basic event 42 that has a causal relationship with the top event 40 or the intermediate event 41 in accordance with a pre-created processing model.

[0100] According to the embodiment, the maintenance work support system 1 can quickly restore the industrial machine 2 to a normal state while improving the work efficiency of the maintenance company that responds to the occurrence or symptoms of an abnormality in the industrial machine 2. The cause of the malfunction can be accurately identified by the confirmation work that utilizes the history information.

[0101] The maintenance work support system 1 includes a resource inference engine 24 that infers the resources required for the work to resolve the basic event 42 identified by the basic event identification engine 23, in accordance with a resource inference model 50 created in advance. As a result, the resources required for the work to resolve the defect are determined from the identified basic event 42.

[0102] The maintenance work support system 1 includes an item procurement application 26 that checks an item 61 inferred by executing an item inference model 51 against an item management database 32 and procures the item 61, a personnel selection application 27 that checks the capabilities of a worker 60 inferred by executing a personnel inference model 52 against a personnel allocation management database 33 and selects a worker 60 to perform the resolution work, and a content extraction application 28 that checks content 62 inferred by executing a content inference model 53 against a content management database 34 and extracts content 62 to be referenced when performing the resolution work. This makes it possible to arrange for the item 61 required for the resolution work, select a worker 60, and arrange for the content 62 based on the inference information.

[0103] The maintenance work support system 1 includes an action support application 29 that incorporates preparation information about the items 61 procured by the item procurement application 26 and reference information about the content 62 extracted by the content extraction application 28 into daily action plan information for the workers 60 selected by the personnel selection application 27. This allows the selected workers 60 to recognize the items 61 that should be brought to maintenance and the available content 62 simply by checking the daily action plan information.

[0104] The maintenance work support system 1 includes a report management database 35 that accumulates work reports that record the details of the resolution work performed by the workers 60 selected by the personnel selection application 27. This allows the work reports that record the details of the resolution work performed by the workers 60 to be accumulated.

[0105] The maintenance work support system 1 includes a knowledge information extraction engine 25 that extracts knowledge information useful for resolving basic events 42 from the work reports stored in the report management database 35. The knowledge information extraction engine 25 selects cases in which basic events 42 have been successfully resolved from the work reports, and uses a pre-created language model to read characteristic text contained in the records of the successful cases, performs natural language processing, and outputs the generated sentences as knowledge information. This makes it possible to sequentially accumulate sentences of cases in which basic events 42 have been successfully resolved.

[0106] The maintenance work support system 1 includes a model update application 30 that updates at least one of the causal model in the basic event inference engine 20, the processing model in the confirmation work presentation engine 21, the processing model in the basic event identification engine 23, and the resource inference model in the resource inference engine 24 by inputting knowledge information extracted by the knowledge information extraction engine 25. This updates the model (norm) to one that reflects the knowledge information.

[0107] Furthermore, in a maintenance work operation method that utilizes the maintenance work support system 1 to run a PDCA (Plan, Do, Check, Act) cycle, the planning process is a process involving the identification of basic events using a basic event inference engine 20, a confirmation work presentation engine 21, and a basic event identification engine 23, the execution process is a process involving the execution of resolution work using a resource inference engine 24 and an action support application 29, the evaluation process is a process involving the extraction of useful knowledge information using a knowledge information extraction engine 25, and the improvement process is a process involving model updating using a model update application 30. In this way, by continuing to run the PDCA cycle, the efficiency of maintenance work will be dramatically improved.

[0108] [Other embodiments] In the above-described embodiment, the information processing device 6 functions as multiple engines and applications, and the server 5 functions as a database. At least some of the functions of the server 5 may be provided in the information processing device 6, or at least some of the functions of the information processing device 6 may be provided in the server 5. The server 5 and the information processing device 6 may be configured by a single computer 10. The multiple engines, multiple applications, and multiple databases may be configured by separate hardware or software.

[0109] <Contribution to the United Nations-led Sustainable Development Goals (SDGs)> The maintenance work support system disclosed herein can improve the operational efficiency of maintenance companies responding to abnormalities and signs of abnormalities in industrial machinery and equipment, while quickly restoring the equipment to normal operation. This can prevent a decline in the energy efficiency of industrial machinery and equipment, contributing to the achievement of Goal 7 of the Sustainable Development Goals (SDGs), "Affordable and clean energy." Furthermore, maintaining energy efficiency can also curb increases in carbon dioxide emissions, contributing to the achievement of Goal 13, "Take urgent action to combat climate change." [Explanation of symbols]

[0110] 1...Maintenance work support system, 2...Industrial machinery, 3...Business establishment, 4...Environmental sensor, 5...Server, 6...Information processing device, 7...Information terminal, 8...Input device, 9...Display device, 10...Computer, 11...Processor, 12...Storage device, 13...Communication interface, 14...Input / output interface, 15...Application software, 20...Basic event inference engine, 21...Confirmation work presentation engine, 22...History information extraction engine, 23...Basic event identification engine, 24...Resource inference engine, 25...Knowledge information extraction engine, 26...Item procurement application, 27...Personnel selection application, 28...Content extraction application, 29...Action support application, 30...Model update application, 31...History management database, 32... Item management database, 33...staffing management database, 34...content management database, 35...report management database, 40...top event, 41...intermediate event, 42...basic event, 43...non-expansion event, 44...AND gate, 45...OR gate, 50...resource inference model, 51...item inference model, 52...personnel inference model, 53...content inference model, 60...worker, 61...item, 62...content, 90...portal screen, 91...first text box, 91a...execute button, 92...first window, 93...second window, 94...third window, 95...fourth window, 96...second text box, 96a...search button, 97...fifth window, 98...result input box, 98a...decision button, 99...sixth window.

Claims

1. a basic event inference engine that infers one or more basic events that cause the malfunction and their occurrence probability in accordance with a causal model created in advance by inputting malfunction event information related to a top event or intermediate event related to a malfunction of industrial machinery; a confirmation task presentation engine that, by inputting the basic events inferred by the basic event inference engine and their occurrence probabilities, presents confirmation tasks for confirming whether or not the basic events have occurred in accordance with a pre-created processing model in descending order of the probability of the basic events occurring; a history information extraction engine that extracts, by inputting identification information of the industrial machine in which a malfunction has occurred, history information that can be used for confirmation work, including the history of environmental information related to the operating environment of the target industrial machine and the history of maintenance work records from a history management database; a basic event identification engine that identifies a basic event having a causal relationship with a top event or an intermediate event according to a pre-created processing model by inputting the results of the required confirmation work presented by the confirmation work presentation engine; Maintenance work support system.

2. a resource inference engine for inferring resources required for a resolution task of the basic event identified by the basic event identification engine in accordance with a resource inference model created in advance; The maintenance work support system according to claim 1 .

3. The resource inference model includes at least an item inference model for inferring items required for the work to resolve the basic event, a human resources inference model for inferring the capabilities of workers required for the work to resolve the basic event, and a content inference model for inferring content useful for the work to resolve the basic event. The maintenance work support system according to claim 2 .

4. an item procurement application that checks the items inferred by executing the item inference model against an item management database to procure the items; a personnel selection application that compares the capabilities of the workers inferred by executing the personnel inference model with a personnel allocation management database to select workers to perform the resolution work; a content extraction application that compares the content inferred by executing the content inference model with a content management database to extract content to be referenced when performing a resolution operation; The maintenance work support system according to claim 3 .

5. an action support application that incorporates preparation information about the items procured by the item procurement application and reference information about the content extracted by the content extraction application into daily action plan information for the workers selected by the personnel selection application; The maintenance work support system according to claim 4.

6. a report management database that stores work reports recording the details of the resolution work performed by the workers selected by the personnel selection application; The maintenance work support system according to any one of claims 2 to 5.

7. It is equipped with a knowledge information extraction engine that extracts knowledge information useful for resolving basic issues from work reports stored in the report management database. The knowledge information extraction engine Select cases from the work reports that have successfully resolved basic issues, Using a pre-created language model, the system reads characteristic text contained in records of success cases, performs natural language processing, and outputs the generated sentences as knowledge information. The maintenance work support system according to claim 6.

8. a model update application that updates at least one of a causal model in the basic event inference engine, a processing model in the confirmation task presentation engine, a processing model in the basic event identification engine, and a resource inference model in the resource inference engine by inputting knowledge information extracted by the knowledge information extraction engine; The maintenance work support system according to claim 7.

9. A maintenance work operation method for implementing a PDCA (Plan, Do, Check, Check, Action) cycle by utilizing the maintenance work support system according to claim 8, The planning process is a process involving identification of basic events using a basic event inference engine, a confirmation task presentation engine, and a basic event identification engine; The execution process involves the performance of a resolution task using a resource inference engine and a behavior support application; The evaluation process involves extracting useful knowledge information using a knowledge information extraction engine, The refinement process involves updating the model using a model update application. Maintenance operation methods.

Images