Intelligent application platform in vertical field of motor train unit and use method

Abstract

The invention discloses a motor train unit vertical field intelligent application platform and a use method, and the platform comprises a large language model, a motor train unit professional knowledge base, a motor train unit professional data set, a motor train unit fault and operation log library, a motor train unit typical fault data set, and an interaction module. The interaction module is in data connection with the big language model through the motor train unit professional knowledge base and the motor train unit fault and operation log base, and the motor train unit professional data set is in data connection with the motor train unit professional knowledge base. The motor train unit typical fault data set is connected with the motor train unit fault and operation log library; according to the application of the large language model in the vertical field of the motor train unit, an automatic and intelligent solution can be provided for fault diagnosis of the motor train unit, new power can be provided for development and innovation of the motor train unit, and innovation and development of the motor train unit are promoted.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent platforms, and particularly to an intelligent application platform for the vertical field of EMUs and a usage method thereof. Background Art

[0002] With the rapid development and continuous expansion of high-speed railways, the types of EMUs are gradually increasing, and the technical and quality requirements for EMUs are also constantly improving. At present, it has become an important development direction in the modern transportation industry. During the operation and maintenance of EMUs, a large amount of professional knowledge and technical support are required.

[0003] Traditional knowledge bases and operation and maintenance systems have limitations in meeting the actual needs of EMUs, unable to provide intelligent assistance and with low query efficiency.

[0004] In view of the above defects, the creator of the present invention finally obtained the present invention through long-term research and practice. Summary of the Invention

[0005] To solve the above technical defects, the technical solution adopted by the present invention is to provide an intelligent application platform for the vertical field of EMUs, including a large language model, an EMU professional knowledge base, an EMU professional data set, an EMU fault and operation log library, an EMU typical fault data set, and an interaction module. The interaction module is respectively connected to the large language model through the EMU professional knowledge base and the EMU fault and operation log library for data connection. The EMU professional data set is connected to the EMU professional knowledge base for data connection, and the EMU typical fault data set is connected to the EMU fault and operation log library.

[0006] Preferably, the EMU professional data set is classified by data type, vehicle type, vehicle part, and operation process.

[0007] Preferably, the EMU typical fault data set is classified by data type, vehicle type, fault part, and fault type.

[0008] Preferably, the large language model, the interaction module are connected to the EMU professional data set for data connection.

[0009] Preferably, a usage method of the intelligent application platform for the vertical field of EMUs includes the steps of:

[0010] S1, input the input information through the interaction module and preprocess the input information;

[0011] S2, post-process the input information through the large language model to obtain output information, and output the output information through the interaction module.

[0012] Preferably, in the step S1, the preprocessing process is as follows: when the input information is determined to be related to the EMU knowledge Q&A judgment, the input information is processed by the large language model via the EMU professional knowledge base; when the input information is determined not to be related to the EMU knowledge Q&A judgment, it is determined whether it is related to the EMU fault diagnosis and detection; when the input information is determined to be related to the EMU fault diagnosis and detection, the input information is processed by the large language model via the EMU fault and operation log library; when the input information is determined not to be related to the EMU fault diagnosis and detection, the input information is directly input into the large language model as fine-tuning information for realizing the fine-tuning training of the basic knowledge of the EMU of the large language model.

[0013] Preferably, when the input information includes both the part related to the EMU knowledge Q&A judgment and the part related to the EMU fault diagnosis and detection, the part related to the EMU knowledge Q&A judgment is paired and integrated by the EMU professional knowledge base, and the part related to the EMU fault diagnosis and detection is paired and integrated by the EMU fault and operation log library, and both are sorted by the vector library and then handed over to the large language model for post-processing.

[0014] Preferably, based on the input information, the large language model analyzes and sorts in cooperation with the EMU professional knowledge base and the EMU fault and operation log library, and converts it into the output information in text format, which is output by the interaction module.

[0015] Preferably, the large language model automatically typesets, intelligently corrects errors, and polishes the text of the output information in text format.

[0016] Preferably, the output information after post-processing by the large language model is optimized, verified, and screened and incorporated into the EMU professional data set to update the EMU professional data set.

[0017] Compared with the prior art, the beneficial effects of the present invention are as follows: the application of the large language model in the vertical field of EMUs in the present invention can provide an automated and intelligent solution for EMU fault diagnosis, can provide new impetus for the development and innovation of EMUs, and promote the innovation and development of EMUs. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a schematic diagram of the architecture of the intelligent application platform for the vertical field of EMUs. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0019] The following further describes the above and other technical features and advantages of the present invention in more detail with reference to the accompanying drawings.

[0020] Embodiment 1

[0021] As shown Figure 1 in Figure 1 the figure, it is a schematic diagram of the architecture of the intelligent application platform for the vertical field of the multiple unit train.

[0022] The intelligent application platform for the vertical field of the multiple unit train described in the present invention includes a large language model, a multiple unit train professional knowledge base, a multiple unit train professional data set, a multiple unit train fault and operation log library, a multiple unit train typical fault data set, and an interaction module. The interaction module is respectively connected to the large language model through the multiple unit train professional knowledge base and the multiple unit train fault and operation log library for data connection. The multiple unit train professional data set is connected to the multiple unit train professional knowledge base for data connection. The multiple unit train typical fault data set is connected to the multiple unit train fault and operation log library.

[0023] The data in the multiple unit train professional data set is classified by data type, vehicle type, vehicle part, and operation process, which is convenient for the multiple unit train professional knowledge base to screen and retrieve the data in the multiple unit train professional data set.

[0024] The data in the multiple unit train typical fault data set is classified by data type, vehicle type, fault part, and fault type, which is convenient for the multiple unit train fault and operation log library to screen and retrieve the data in the multiple unit train typical fault data set.

[0025] Structured processing is performed on documents such as operation manuals, rules and regulations, basic knowledge, typical fault cases, and equipment logs related to multiple unit trains. Corresponding knowledge bases, namely the multiple unit train professional knowledge base and the multiple unit train fault and operation log library, are established for different vehicle types (such as CRH1A, CRH1E, CRH3C, CR400AF, etc.) to support the professional Q&A and fault diagnosis functions of the platform. According to the operation manual and the typical fault case library, the fault cause can be quickly analyzed and the emergency fault handling process can be given. Through the online data interaction function, technicians can view and analyze the status data of the multiple unit train in real time, and discover and handle faults in time.

[0026] By learning a large number of emergency fault handling and typical fault cases in the multiple unit train typical fault data set, the large language model is enabled to have the ability to intelligently judge the fault point.

[0027] The intelligent application platform for the vertical field of the multiple unit train can also generate simulated fault points for practice based on the multiple unit train fault and operation log library, improving the efficiency and accuracy of fault handling. At the same time, it has capabilities such as text-to-sql, realizing online data interaction, and facilitating data analysis by technicians.

[0028] The large language model, the interaction module, and the EMU professional dataset are data-connected. The results processed by the large language model are optimized, verified, and screened and then incorporated into the EMU professional dataset, continuously improving the large language model's ability in EMU professional knowledge. At the same time, the present invention supports the continuous update and sharing functions of knowledge. Technical personnel can upload new knowledge and experience to the platform at any time for others to learn and reference. This helps to promote the innovation and development of EMU technology and improve the competitiveness of the entire industry.

[0029] By pre-training and fine-tuning the large language model, it masters the basic knowledge in the EMU professional field and has the ability of professional Q&A. Technical and maintenance personnel can quickly obtain accurate technical support and guidance through the platform.

[0030] The present invention is used to build an EMU intelligent platform integrating knowledge Q&A, document writing, fault diagnosis and monitoring. The platform has a friendly user interface and efficient data processing capabilities to meet the needs of technical personnel in actual work.

[0031] The application of the large language model in the EMU vertical field in the present invention can provide automated and intelligent solutions for EMU fault diagnosis, can provide new impetus for the development and innovation of EMUs, and promote the innovation and development of EMUs.

[0032] Embodiment 2

[0033] The usage method of the EMU vertical field intelligent application platform of the present invention includes the steps:

[0034] S1. Input the input information through the interaction module and preprocess the input information.

[0035] S2. Post-process the input information through the large language model to obtain output information, and output the output information through the interaction module.

[0036] Specifically, in step S1, the process of the preprocessing is as follows: when the input information is judged to be relevant to EMU knowledge Q&A judgment, the input information is processed by the large language model via the EMU professional knowledge base; when the input information is judged not to be relevant to EMU knowledge Q&A judgment, it is judged whether it is relevant to EMU fault diagnosis and detection; when the input information is judged to be relevant to EMU fault diagnosis and detection, the input information is processed by the large language model via the EMU fault and operation log library; when the input information is judged not to be relevant to EMU fault diagnosis and detection, the input information is directly input into the large language model as fine-tuning information for realizing the fine-tuning training of the basic knowledge of EMUs of the large language model.

[0037] When the input information includes both the relevant part of the EMU knowledge Q&A judgment and the relevant part of the EMU fault diagnosis and detection, the relevant part of the EMU knowledge Q&A judgment is paired and integrated by the EMU professional knowledge base, and the relevant part of the EMU fault diagnosis and detection is paired and integrated by the EMU fault and operation log library. After being sorted by the vector library, both are handed over to the large language model for post-processing.

[0038] Based on the input information, the large language model analyzes and sorts in cooperation with the EMU professional knowledge base and the EMU fault and operation log library, and converts it into the output information in text format, which is output by the interaction module. Thus, the present invention is used to write documents such as summaries, monthly summaries, annual summaries, analysis reports, papers, etc., improving work efficiency and quality; the large language model enables the present invention to write professional analysis reports, papers, etc. with complex structures, and the large language model also provides functions such as automatic typesetting, intelligent error correction, and text polishing to improve the quality and efficiency of the documents.

[0039] Preferably, the output information after post-processing by the large language model is optimized, verified, and screened and incorporated into the EMU professional data set to update the EMU professional data set, thereby optimizing the EMU professional knowledge base to continuously improve the large language model's ability to handle EMU professional knowledge.

[0040] The above are only the preferred embodiments of the present invention, which are illustrative rather than restrictive to the present invention. Those skilled in the art understand that many changes, modifications, and even equivalents can be made within the spirit and scope defined by the claims of the present invention, but all will fall within the protection scope of the present invention.

Claims

1. An intelligent application platform for the vertical field of EMUs, characterized in that: It includes a large language model, an EMU professional knowledge base, an EMU professional data set, an EMU fault and operation log library, an EMU typical fault data set and an interaction module. The interaction module is respectively connected to the large language model through the EMU professional knowledge base and the EMU fault and operation log library, the EMU professional data set is connected to the EMU professional knowledge base, and the EMU typical fault data set is connected to the EMU fault and operation log library.

2. The vertical intelligent application platform for EMUs according to claim 1, characterized in that: The EMU professional data set is classified by data type, vehicle model, vehicle part and operation process.

3. The EMU vertical field intelligent application platform as claimed in claim 2, characterized in that: The EMU typical fault data set is classified by data type, vehicle model, fault location and fault type.

4. The vertical intelligent application platform for EMUs as claimed in claim 3, characterized in that: The large language model, the interaction module and the EMU professional data set are data connected.

5. A method for using the EMU vertical field intelligent application platform as claimed in claim 4, characterized in that: Includes steps: S1, inputting input information through the interaction module and preprocessing the input information; S2, post-processing the input information through the large language model to obtain output information, and outputting the output information through the interaction module.

6. The method of use according to claim 5, characterized in that: In the step S1, the preprocessing process is: when the input information is judged to be relevant to the EMU knowledge question and answer judgment, the input information is processed via the EMU professional knowledge base to the large language model; when the input information is judged not to be relevant to the EMU knowledge question and answer judgment, it is judged whether it is related to the EMU fault diagnosis and detection; when the input information is judged to be related to the EMU fault diagnosis and detection, the input information is processed via the EMU fault and operation log library to the large language model; when the input information is judged not to be related to the EMU fault diagnosis and detection, the input information is directly input into the large language model as fine-tuning information, so as to realize the EMU basic knowledge fine-tuning training of the large language model.

7. The method of use according to claim 5, characterized in that: When the input information includes both the EMU knowledge question and answer judgment related part and the EMU fault diagnosis and detection related part, the EMU knowledge question and answer judgment related part is paired and integrated by the EMU professional knowledge base, and the EMU fault diagnosis and detection related part is paired and integrated by the EMU fault and operation log base, and both are sorted by the vector library and then handed over to the large language model for post-processing.

8. The method of use according to claim 5, characterized in that: The large language model analyzes and organizes the input information based on the input information in cooperation with the EMU professional knowledge base and the EMU fault and operation log base, and converts the output information into a text format, which is then output by the interactive module.

9. The method of use according to claim 8, characterized in that: The large language model automatically typesets, intelligently corrects errors, and polishes the output information in text format.

10. The method of use according to claim 5, characterized in that: The output information after post-processing by the large language model is optimized, verified, screened and incorporated into the EMU professional data set to update the EMU professional data set.

Images