Intelligent operation and maintenance method for urban rail PHM system based on large language model

Abstract

The invention relates to an urban rail PHM system-oriented intelligent operation and maintenance method based on a large language model, and the method comprises the steps: obtaining historical work order data through an urban rail PHM system, constructing a data set through the data which successfully completes the operation and maintenance, and training a large language model, thereby obtaining a fault analysis model capable of carrying out the fault prediction and analysis. And after the fault occurs, the fault condition data is imported into the fault analysis model to obtain a corresponding fault analysis result and a fault disposal scheme, so that a basis for overhauling can be quickly provided for operation and maintenance personnel, and the operation and maintenance overhauling cost is reduced while the operation and maintenance overhauling efficiency of the railway vehicle is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of rail vehicle operation and maintenance, and in particular to an intelligent operation and maintenance method for an urban rail PHM system based on a large language model. Background Art

[0002] The PHM system, whose full name is Prognostics Health Management, is called Fault Prediction and Health Management System in Chinese. It is used to receive real-time or offline data from the train network or other subsystems, establish a vehicle digital twin model, and after a series of comprehensive diagnostic analysis, display the analysis results to users in the form of various report graphics or information push. It gives recommended treatment measures for diagnosed vehicle faults, provides necessary support for vehicle fault repair, and achieves the goal of rapid decision-making assistance and operation management support.

[0003] As intelligent operation and maintenance systems, led by PHM, are gradually applied to new and renovated urban rail transit projects, actual operation data and inspection data are stored with high quality, but the accumulation of data has not brought significant improvements to daily maintenance and inspection work. According to statistics, the man-to-vehicle ratio required for projects equipped with intelligent operation and maintenance systems is less than 0.1 lower than that of conventional projects. In some projects, the man-to-vehicle ratio has increased due to the need to equip intelligent operation and maintenance personnel.

[0004] Therefore, there is an increasing demand to develop models with strong generalization capabilities based on the actual operation and maintenance data of the PHM system and the integrated expert knowledge base, which can quickly locate faults and handle various types of tasks. Summary of the invention

[0005] A main purpose of the present invention is to achieve at least one of the above purposes, that is, to provide an intelligent operation and maintenance method for an urban rail PHM system based on a large language model, which can improve the operation and maintenance efficiency of rail vehicles while reducing the operation and maintenance costs.

[0006] In order to achieve the above object, the technical solution adopted by the present invention is: The present invention provides an intelligent operation and maintenance method for an urban rail PHM system based on a large language model, which includes the following working steps: Acquiring and preprocessing raw data: acquiring several historical work order data of urban rail operation and maintenance through the urban rail PHM system to form raw data, and preprocessing the raw data. After the preprocessing is completed, the raw data is divided into a training data set and a test data set according to a proportion; Model training: the large language model is trained by the training data set to obtain a pre-trained model, the pre-trained model obtained is evaluated by the test data set, and the pre-trained model obtained by the previous training is re-trained and evaluated by the training data set after adjusting the learning rate, and the step of re-training and evaluating is repeated until the evaluation is qualified, then the model training is completed, and the trained model is output as a fault analysis model; Operation and maintenance analysis: Import fault condition data and input it into the fault analysis model, and output fault analysis results and fault handling solutions through the fault analysis model.

[0007] According to one embodiment of the present invention, each historical work order data includes multiple of the fault component name, fault system, fault subsystem, fault subject, fault phenomenon, fault location picture, fault description, report level, fault handling measures and handling measure description.

[0008] According to one of the embodiments of the present invention, during model training, the test data set is evaluated in combination with manual confirmation and adjustment to improve the working accuracy of the final fault analysis model.

[0009] According to one embodiment of the present invention, the fault condition data includes at least one of a fault scene description and a fault scene picture.

[0010] According to one implementation manner of the present invention, the fault analysis result includes fault location and fault cause analysis.

[0011] According to one embodiment of the present invention, if there are multiple fault analysis results that match the input fault condition data, each fault analysis result is output, and the matching probability corresponding to each fault analysis result is output at the same time.

[0012] According to one embodiment of the present invention, when performing operation and maintenance analysis, the fault condition data is imported manually or automatically by the urban rail PHM system.

[0013] According to one embodiment of the present invention, when performing operation and maintenance analysis, the output fault analysis results and fault handling solutions are fed back and added to the fault analysis model after being confirmed accurate by operation and maintenance personnel during on-site maintenance, so as to improve the fault analysis model.

[0014] In particular, the present invention also provides a computing device, comprising a memory, a processor, and a computer program stored in the memory and executable by the processor, wherein the processor implements the method as described above when executing the computer program.

[0015] The present invention also provides a computer-readable storage medium, in which a computer program is stored, and when the computer program is executed by a processor, the method described above is implemented. The computer-readable storage medium is preferably a non-volatile readable storage medium.

[0016] Compared with the prior art, the advantages and beneficial effects of the intelligent operation and maintenance method for the urban rail PHM system based on the large language model in the patent application of the present invention are: The intelligent operation and maintenance method for the urban rail PHM system based on a large language model of the present application obtains historical work order data through the urban rail PHM system, and constructs a data set through the data of successfully completed operation and maintenance inspections in the past to train the large language model, thereby obtaining a fault analysis model that can perform fault prediction analysis. After a fault occurs, the fault condition data is imported into the fault analysis model to obtain the corresponding fault analysis results and fault handling solutions, which can quickly provide the operation and maintenance personnel with a basis for maintenance, thereby improving the operation and maintenance efficiency of rail vehicles while reducing the operation and maintenance costs.

[0017] Furthermore, this application studies the PHM system for intelligent operation and maintenance to obtain a general large language model for troubleshooting guidelines, which can greatly reduce the complexity and cost of developing and maintaining different models and improve the efficiency of troubleshooting and handling. Since the large language model has learned a lot of knowledge related to rail transit and intelligent operation and maintenance in the pre-training stage to train the fault analysis model, it only needs to fine-tune the operation and maintenance personnel to standardize the fault situation data to make the fault analysis model adapt to the characteristics of intelligent operation and maintenance, and can have outstanding performance in the field of rail transit intelligent operation and maintenance where data is scarce. The intelligent operation and maintenance method of this application aims to provide practical guidelines for intelligent operation and maintenance and maintenance management based on the experience of the expert knowledge base, thereby reducing operation and maintenance costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Hereinafter, some specific embodiments of the present invention will be described in detail in an exemplary and non-limiting manner with reference to the accompanying drawings. The same reference numerals in the accompanying drawings indicate the same or similar components or parts. It should be understood by those skilled in the art that these drawings are not necessarily drawn to scale. In the accompanying drawings: Figure 1 is a schematic diagram of the overall workflow of an intelligent operation and maintenance method according to an embodiment of the present invention; Figure 2 is a schematic diagram of a process for performing operation and maintenance by manually acquiring fault condition data according to an embodiment of the present invention; Figure 3 It is a diagram showing the results of operation and maintenance after manually acquiring fault condition data according to an embodiment of the present invention; Figure 4is a schematic diagram of a process for automatically acquiring fault condition data for operation and maintenance through an urban rail PHM system according to another embodiment of the present invention; Figure 5 is a diagram showing the results of operation and maintenance after automatically acquiring fault condition data through the urban rail PHM system according to another embodiment of the present invention; Figure 6 is a structural diagram of a computing device according to another embodiment of the present application; Figure 7 It is a structural diagram of a computer-readable storage medium according to another embodiment of the present application. DETAILED DESCRIPTION

[0019] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0020] In addition, the technical features involved in the different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0021] This embodiment describes an intelligent operation and maintenance method for urban rail PHM system based on a large language model. Figure 1 As shown, it includes the following working steps: Acquiring and preprocessing raw data: acquiring several historical work order data of urban rail operation and maintenance through the urban rail PHM system to form raw data, and preprocessing the raw data. After the preprocessing is completed, the raw data is divided into a training data set and a test data set according to a proportion; Model training: the large language model is trained by the training data set to obtain a pre-trained model, the pre-trained model obtained is evaluated by the test data set, and the pre-trained model obtained by the previous training is re-trained and evaluated by the training data set after adjusting the learning rate, and the step of re-training and evaluating is repeated until the evaluation is qualified, then the model training is completed, and the trained model is output as a fault analysis model; Operation and maintenance analysis: Import fault condition data and input it into the fault analysis model, and output the fault analysis results and fault handling plan through the fault analysis model. The fault condition data includes at least one of the fault site description and fault site picture, which can be easily obtained manually by operation and maintenance personnel or captured by the urban rail PHM system through the train working status data. The fault analysis results include fault location and fault cause analysis, which can accurately reflect the location and cause of the fault.

[0022] It is not difficult to understand that the intelligent operation and maintenance method of this embodiment is to obtain historical work order data through the urban rail PHM system, and to build a data set through these data of successfully completed operation and maintenance inspections in the past to train the large language model, so as to obtain a fault analysis model that can perform fault prediction analysis. After a fault occurs, the fault condition data is imported into the fault analysis model to obtain the corresponding fault analysis results and fault handling plans. This can quickly provide the operation and maintenance personnel with a basis for maintenance, thereby improving the operation and maintenance efficiency of rail vehicles while reducing the operation and maintenance costs.

[0023] The process of preprocessing the raw data can generally include data cleaning actions such as data completion and filling, and invalid data removal. Each historical work order data includes multiple of the fault part name, fault system, fault subsystem, fault subject, fault phenomenon, fault part picture, fault description, report level, fault handling measures and handling measures description. If the historical work order data obtained from the urban rail PHM system lacks corresponding parameters, it will be screened out and required to be supplemented by the user, or directly treated as invalid data and not involved in the model training process.

[0024] The training of the fault analysis model adopts the offline deployment method. In this embodiment, two basic models, iFlytek Spark Pro and llama-3-chinese-8b-instruct, are used for training. The historical work order data is divided into 80% training set and 20% test set, the learning rate is 0.0001, and the number of training times is 5. During model training, the test data set is evaluated and manually confirmed and adjusted, that is, it is manually confirmed by specific relevant engineers to improve the working accuracy of the final fault analysis model and ensure that the result data generated by the large model is true and reliable.

[0025] The intelligent operation and maintenance method of the present application can have two working modes when performing operation and maintenance of rail trains: The first is the manual maintenance mode, where the operation and maintenance personnel manually discover the fault, enter the fault scene description and the fault scene pictures taken to the upload page of the fault analysis model. The fault analysis model has strong analysis capabilities after being trained based on the data set composed of historical work order data, and can generate corresponding fault locations (such as Figure 3 As shown in the figure), fault cause analysis, and corresponding fault handling measures; the specific work flow is as follows Figure 2 After completing this troubleshooting work, the fault analysis model will enter the relevant data into the database to achieve the effect of continuously improving the model itself.

[0026] The second is the automatic maintenance mode. After the urban rail PHM system automatically detects a fault, it inputs the fault condition data (such as fault definition) into the fault analysis model. The fault analysis model has strong analytical capabilities after being trained based on the data set composed of historical work order data. It can generate corresponding fault location, fault cause analysis, and corresponding fault handling measures, forming a Figure 5 The fault report shown in the figure; the specific workflow is as follows Figure 4 After completing this troubleshooting work, the fault analysis model will enter the relevant data into the database to achieve the effect of continuously improving the model itself.

[0027] Preferably, when the operation and maintenance analysis is performed, the output fault analysis results and fault handling solutions are fed back and added to the fault analysis model after being confirmed accurate by the operation and maintenance personnel during on-site maintenance, so as to improve the fault analysis model.

[0028] In addition, there may be multiple results obtained from the fault analysis model that are compatible with the input fault condition data, because different faults may produce the same fault manifestation. If there are multiple fault analysis results that match the input fault condition data, each fault analysis result will be output, and the matching probability corresponding to each fault analysis result will be output at the same time.

[0029] In summary, this application studies the PHM system for intelligent operation and maintenance to obtain a general large language model for troubleshooting guidelines, which can greatly reduce the complexity and cost of developing and maintaining different models and improve the efficiency of troubleshooting and handling. Since the large language model has learned a lot of knowledge related to rail transit and intelligent operation and maintenance in the pre-training stage to train the fault analysis model, it only needs to fine-tune the operation and maintenance personnel to standardize the fault situation data to make the fault analysis model adapt to the characteristics of intelligent operation and maintenance, and can have outstanding performance in the field of rail transit intelligent operation and maintenance where data is scarce. The intelligent operation and maintenance method of this application aims to provide practical guidelines for intelligent operation and maintenance and maintenance management based on the experience of the expert knowledge base, thereby reducing the cost of operation and maintenance.

[0030] Based on the detailed description of the specific embodiments of the present application in combination with the accompanying drawings below, those skilled in the art will become more aware of the above and other objects, advantages and features of the present application.

[0031] The present application also provides a computing device, referring to Figure 6 The computing device includes a memory 1120, a processor 1110, and a computer program stored in the memory 1120 and executable by the processor 1110. The computer program is stored in a space 1130 for program code in the memory 1120. When the computer program is executed by the processor 1110, it is implemented to execute any one of the method steps 1131 according to the present invention.

[0032] The present application also provides a computer-readable storage medium. Figure 7 The computer-readable storage medium comprises a storage unit for program code, the storage unit being provided with a program 1131 ′ for executing the method steps according to the present invention, the program being executed by a processor.

[0033] The embodiment of the present application further provides a computer program product including instructions. When the computer program product is run on a computer, the computer is caused to execute the method steps according to the present invention.

[0034] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer loads and executes the computer program instructions, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from a website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer, server or data center. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more available media integrated. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state drive Solid State Disk (SSD)), etc.

[0035] The professionals should also be further aware that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented with electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0036] A person skilled in the art can understand that all or part of the steps in the above-mentioned embodiment method can be completed by instructing a processor through a program, and the program can be stored in a computer-readable storage medium, and the storage medium is a non-transitory medium, such as a random access memory, a read-only memory, a flash memory, a hard disk, a solid-state drive, a magnetic tape, a floppy disk, an optical disc, and any combination thereof.

[0037] In particular, the present invention also provides a computing device, comprising a memory, a processor, and a computer program stored in the memory and executable by the processor, wherein the processor implements the method as described above when executing the computer program.

[0038] The present invention also provides a computer-readable storage medium, in which a computer program is stored, and when the computer program is executed by a processor, the method described above is implemented. The computer-readable storage medium is preferably a non-volatile readable storage medium.

[0039] The above embodiments are only for illustrating the technical concept and features of the present invention, and their purpose is to enable people familiar with this technology to understand the contents of the present invention and implement them accordingly. They cannot be used to limit the protection scope of the present invention. Any equivalent changes or modifications made according to the spirit of the present invention should be included in the protection scope of the present invention.

Claims

1. An intelligent operation and maintenance method for urban rail PHM system based on a large language model, characterized in that: The following steps are included: Acquiring and preprocessing raw data: acquiring several historical work order data of urban rail operation and maintenance through the urban rail PHM system to form raw data, and preprocessing the raw data. After the preprocessing is completed, the raw data is divided into a training data set and a test data set according to a proportion; Model training: the large language model is trained by the training data set to obtain a pre-trained model, the pre-trained model obtained is evaluated by the test data set, and the pre-trained model obtained by the previous training is re-trained and evaluated by the training data set after adjusting the learning rate, and the step of re-training and evaluating is repeated until the evaluation is qualified, then the model training is completed, and the trained model is output as a fault analysis model; Operation and maintenance analysis: Import fault condition data and input it into the fault analysis model, and output fault analysis results and fault handling solutions through the fault analysis model.

2. The intelligent operation and maintenance method for urban rail PHM system based on large language model according to claim 1 is characterized in that: Each historical work order data includes multiple of the fault component name, fault system, fault subsystem, fault subject, fault phenomenon, fault location picture, fault description, report level, fault handling measures and handling measure description.

3. The intelligent operation and maintenance method for urban rail PHM system based on large language model according to claim 1 is characterized in that: During model training, the test data set is evaluated in combination with manual confirmation and adjustment to improve the working accuracy of the final fault analysis model.

4. The intelligent operation and maintenance method for urban rail PHM system based on a large language model according to any one of claims 1 to 3, characterized in that: The fault condition data includes at least one of a fault site description and a fault site picture.

5. The intelligent operation and maintenance method for urban rail PHM system based on large language model according to claim 4 is characterized in that: The fault analysis results include fault location and fault cause analysis.

6. The intelligent operation and maintenance method for urban rail PHM system based on large language model according to claim 5 is characterized in that: If there are multiple fault analysis results that match the input fault condition data, each fault analysis result is output, and the matching probability corresponding to each fault analysis result is output at the same time.

7. The intelligent operation and maintenance method for urban rail PHM system based on large language model according to claim 1 is characterized in that: When performing operation and maintenance analysis, the fault condition data is imported manually or automatically by the urban rail PHM system.

8. The intelligent operation and maintenance method for urban rail PHM system based on large language model according to claim 1 is characterized in that: When performing operation and maintenance analysis, the output fault analysis results and fault handling solutions are fed back and added to the fault analysis model after being confirmed accurate by the operation and maintenance personnel during on-site maintenance, so as to improve the fault analysis model.

9. A computing device comprising a memory, a processor, and a computer program stored in the memory and executable by the processor, wherein: When the processor executes the computer program, the method according to any one of claims 1 to 8 is implemented.

10. A computer-readable storage medium having a computer program stored therein, wherein the computer program implements the method according to any one of claims 1 to 8 when executed by a processor.

Images