Rail transit intelligent operation and maintenance system and method based on edge computing and machine learning

Abstract

The application discloses a rail transit intelligent operation and maintenance system and method based on edge computing and machine learning, and the system comprises an end service layer, an edge service layer and a cloud service layer; the end service layer comprises an end sensor, an end controller, an end server and a vehicle TCMS system; the edge service layer is located and comprises a line intelligent operation and maintenance system, a production business management system and a vehicle intelligent maintenance system; and the cloud service layer comprises an intelligent operation and maintenance big data center.The application fully utilizes the Internet of Things technology, realizes unified data collection on the vehicle, and enables the data to be deeply fused; the data can be timely and locally processed; machine algorithms are utilized to realize optimization, which improves the safety of the vehicle system and also provides data decision support for operation and maintenance personnel; through end-end cooperation, the end server on the vehicle can be connected and communicated with the end servers of the vehicles within a set distance through self wireless communication, can intelligently perceive the health status and fault information of the vehicles within the set distance in real time, and can timely avoid safety hazards.

Description

Technical Field

[0001] This invention relates to the field of rail transit operation and maintenance technology, specifically to an intelligent rail transit operation and maintenance system and method based on edge computing and machine learning. Background Technology

[0002] With the development of the rail transit industry, the number of rail vehicles put into operation is increasing year by year. As rail vehicles develop towards intelligence, real-time diagnostics are typically required to accurately assess their health status in order to ensure safe operation. In the traditional model, multiple disciplines involved in rail transit operation, including rolling stock, power supply, signaling, platform screen doors, and track maintenance, each deploy their own intelligent operation and maintenance system. This system utilizes advanced sensor technology to acquire real-time operational status information of the managed system and leverages advanced technologies such as big data and artificial intelligence, along with fuzzy logic and other reasoning algorithms, to assess the health status and predict faults based on real-time data, current operating conditions, and historical data. Making decisions based on manually reading the operational status of each system is neither timely nor efficient. Furthermore, the lack of timely and effective analysis and data sharing among the systems creates "data silos," leading to insufficient overall maintenance coordination. Specifically, the data involved in rail vehicles includes real-time operational status data (doors, bogies, traction power systems, braking systems, onboard signals, auxiliary systems, pantograph-catenary inspection, wheel-rail inspection, etc.), data from intelligent vehicle maintenance systems (intelligent maintenance robots, side running gear inspection, 360° visual inspection, wheelset equivalent conicity inspection, etc.), data from production and business management systems (vehicle status management, platform equipment management, daily operation plan management, construction and maintenance management, key equipment location management, intelligent handheld terminals, etc.), and operation control subsystems. The lack of sufficient information sharing between these systems makes it difficult to accurately locate faults when equipment malfunctions, especially at cross-disciplinary junctions. This results in lengthy troubleshooting times and bottlenecks in specialized data mining.

[0003] First, there are numerous data acquisition devices. This involves a multitude of vehicle on-the-road monitoring data systems, massive amounts of data, inconsistent data interaction protocols, and difficulties in deep data integration, resulting in data silos.

[0004] Secondly, data analysis is not timely. Most of the monitoring data from rail vehicles in transit is not diagnosed or analyzed in real time. For example, vibration data is typically collected over a set period and time, and further analysis and diagnosis are performed after the data collection is completed. This can lead to missed fault points and delays in diagnostic results. Data transmitted wirelessly to a cloud platform for real-time analysis requires high network bandwidth and low latency.

[0005] Thirdly, there is a bottleneck in specialized data mining. Rail transit equipment is diverse, and key subsystem components are highly specialized, resulting in a clear division of labor among technical equipment systems, core components, and key equipment subsystems, requiring highly experienced operation and maintenance personnel. Analyzing and mining data to establish and optimize algorithm models for highly specialized equipment presents significant challenges. These models cannot be updated or self-learned, and using the same, single algorithm model for all vehicles leads to low prediction accuracy and a lack of real-time updates, creating a major bottleneck.

[0006] Fourth, vehicles cannot communicate directly with each other. Currently, vehicles that have not achieved autonomous driving primarily communicate directly with the control center. In extreme situations where there is no network, the vehicle cannot establish effective communication with the outside world, and other vehicles on the same route cannot obtain information about the malfunctioning vehicle, leading to certain safety hazards.

[0007] In conclusion, there is an urgent need for an advanced operation and maintenance method that can improve the safety and efficiency of rail transit vehicle operation and maintenance, and reduce labor costs. Summary of the Invention

[0008] In view of the above-mentioned shortcomings in the prior art, the intelligent operation and maintenance system and method for rail transit based on edge computing and machine learning provided by the present invention solves the problems of low driving safety and low operation and maintenance efficiency of rail transit vehicles in the prior art.

[0009] To achieve the above-mentioned objectives, the technical solution adopted by this invention is as follows:

[0010] A smart rail transit operation and maintenance system based on edge computing and machine learning is provided, which includes an end service layer, an edge service layer and a cloud service layer;

[0011] The end-service layer is located on the vehicle and includes end sensors, end controllers, end servers, and the vehicle TCMS system;

[0012] The edge service layer is located in the ground control center and includes the intelligent line operation and maintenance system, the production business management system, and the intelligent vehicle maintenance system.

[0013] The cloud service layer includes an intelligent operation and maintenance big data center;

[0014] The end sensor is located at the position of the component to be monitored in the vehicle and is used to collect data from the component to be monitored;

[0015] The end controller connects the end sensor and the end server, and is connected to the vehicle TCMS system. It is used to acquire data of the monitored components collected by the end sensor, and to process the collected data in real time through an optimized machine learning algorithm model, and to generate corresponding information and control commands based on the processing results.

[0016] The end server connects the end controller to the intelligent operation and maintenance system of the line, and is connected to the vehicle TCMS system and the end servers of other vehicles within a set distance. It is used to receive the working instructions of the intelligent operation and maintenance system of the line, receive and store the results of the end controller's real-time processing of the collected data through an optimized machine learning algorithm model, generate status information and event information, and upload them to the intelligent operation and maintenance system of the line and the end servers of other vehicles within a set distance.

[0017] The vehicle TCMS system is connected to the end controller to receive corresponding information and control commands generated by the end controller based on the processing results, and to control the vehicle according to the corresponding information and control commands.

[0018] The vehicle TCMS system is connected to the end server to receive the results of the end controller's real-time processing of the collected data through an optimized machine learning algorithm model. Based on the processing results, the system generates corresponding information and control commands and controls the vehicle accordingly.

[0019] The intelligent operation and maintenance system is connected to the end server, the production business management system, the vehicle intelligent maintenance system, and the intelligent operation and maintenance big data center. It is used to receive relevant instructions from the intelligent operation and maintenance big data center, generate corresponding work instructions for the end server, the production business management system, and the vehicle intelligent maintenance system, and upload the status information and event information received from the end server, the production business management system, and the vehicle intelligent maintenance system to the intelligent operation and maintenance big data center.

[0020] The intelligent operation and maintenance big data center is used to aggregate information from vehicles, power supply, signaling, platform screen doors, and track catenary systems, as well as information generated by the production business management system and the vehicle intelligent maintenance system. Based on the diagnostic knowledge base and expert knowledge base, it establishes machine learning algorithm models, receives status and event information through the intelligent line operation and maintenance system, performs health assessments and fault predictions, and generates relevant instructions to be issued to the intelligent line operation and maintenance system. Based on the data received from the end sensors, it updates the diagnostic knowledge base and expert knowledge base through the established machine learning algorithm models, obtaining and issuing updated and optimized machine learning algorithm models.

[0021] Furthermore: Each vehicle includes an end server and at least one end controller, which is located around the component to be monitored, and one end controller monitors at least one end sensor.

[0022] Furthermore, the cloud service layer is deployed in locations including the control center, the owner's group company, and the OEM's operation and maintenance center.

[0023] Furthermore: the intelligent operation and maintenance big data center is connected to at least one line intelligent operation and maintenance system; a line intelligent operation and maintenance system is connected to at least one vehicle; each end server is installed in the corresponding vehicle and is directly connected to or cascaded with the end controller of the corresponding vehicle.

[0024] Furthermore: End sensors include accelerometers, strain sensors, fiber optic sensors, temperature sensors, composite sensors, industrial cameras, millimeter-wave radar, and lidar; Components to be monitored include running gear, pantograph, traction control, doors, air conditioning, and brakes.

[0025] Furthermore, the production business management system includes a vehicle status management module, a platform equipment management module, an operation daily plan management module, a construction and maintenance management module, a key equipment positioning management module, and a smart handheld terminal module.

[0026] Furthermore, the intelligent vehicle inspection system includes an intelligent inspection robot, a vehicle side running gear inspection module, a vehicle 360° vision inspection module, and a wheelset equivalent cone inspection module.

[0027] Furthermore:

[0028] The connection methods between the end sensor and the end controller include wired or wireless connection; the connection methods between the end controller and the end server include wired or wireless connection; the connection methods between the end server and the intelligent operation and maintenance system of the line and the end servers of vehicles within a set distance include wireless connection; the connection methods between the vehicle TCMS system and the end controller or end server include communication cable connection; and the connection methods between the intelligent operation and maintenance system of the line and the intelligent operation and maintenance big data center include wired or wireless connection.

[0029] A method for intelligent operation and maintenance of rail transit based on edge computing and machine learning is provided, which includes the following steps:

[0030] S1. Collect data of the vehicle's monitored components using end sensors and upload it to the end controller;

[0031] S2. Based on the data received from the components to be monitored, the collected data is processed in real time using an optimized machine learning algorithm model, and corresponding information and control commands are generated based on the processing results.

[0032] S3. The end server receives and stores the results of the end controller's real-time processing of the collected data through an optimized machine learning algorithm model, generates status information and event information, and uploads them to the intelligent operation and maintenance system of the line and the end server of other vehicles within a set distance.

[0033] S4. Receive and control the vehicle according to the corresponding information and control instructions generated by the end controller through the vehicle TCMS system; or receive and control the vehicle according to the information and control instructions generated by the end server.

[0034] S5. Receive status and event information from the end server, production business management system and vehicle intelligent maintenance system through the line intelligent operation and maintenance system and upload them to the intelligent operation and maintenance big data center.

[0035] S6. By aggregating system information through the intelligent operation and maintenance big data center, an operation and maintenance knowledge base and an algorithm model library are established; the system information includes information on vehicles, power supply, signaling, platform screen doors, and track catenary, as well as information collected from the production business management system and the vehicle intelligent maintenance system; the operation and maintenance knowledge base includes a diagnostic knowledge base and an expert knowledge base; and the algorithm model library includes machine learning algorithm models.

[0036] S7. The intelligent operation and maintenance big data center performs health assessment and fault prediction on the received status and event information, and generates relevant instructions to be sent to the line intelligent operation and maintenance system; based on the received system data, the diagnostic knowledge base and expert knowledge base are updated through the established machine learning algorithm model, and the updated and optimized machine learning algorithm model is obtained and sent out.

[0037] S8. Receive relevant instructions from the intelligent operation and maintenance big data center through the intelligent operation and maintenance system and generate corresponding work instructions for the end server, production business management system and vehicle intelligent maintenance system.

[0038] S9. The end server, in accordance with the work instructions, cooperates with the end controller, the vehicle TCMS system, and the end servers of other vehicles within a set distance to perform subsequent operation and maintenance work.

[0039] The beneficial effects of this invention are as follows:

[0040] The intelligent operation and maintenance system for rail transit based on edge computing and machine learning addresses the problems of decentralized construction, numerous interfaces, data silos, data transmission and processing delays, inconsistent prediction algorithm models for all vehicles that cannot be updated in a timely manner, and the waste of existing resources caused by the need to build a new system for new functional requirements.

[0041] 1. Fully utilize IoT technology to achieve unified data collection on vehicles, dynamically expand the information collection and hardware capabilities, and deeply integrate data to avoid information silos.

[0042] 2. By utilizing edge computing technology, the end controller of each vehicle can process the collected data in real time through machine learning algorithm models, and generate corresponding vehicle TCMS system information and control commands based on the processing results. This enables the vehicle to process the data collected by the end sensors in real time 24 hours a day and make timely feedback. The data can be processed locally, meeting the requirements of network bandwidth and latency.

[0043] 3. Apply machine learning algorithms to achieve sustainable optimization of vehicle equipment and operating strategies. Under edge-cloud collaboration, the intelligent line maintenance system interacts with the production business management system and the vehicle intelligent maintenance system to obtain relevant data. Centralized big data mining is performed on data related to vehicles and production. Artificial intelligence and machine learning are used to establish various models such as health management, fault early warning, and life prediction to guide the optimization of existing maintenance plans, reduce over-maintenance and under-maintenance of equipment, and avoid operating equipment and vehicles with defects. This improves vehicle system safety and provides data decision support for maintenance personnel. By building a big data maintenance knowledge base, expert data support is provided for on-site maintenance, solving the problem of insufficient skills among maintenance personnel and reducing personnel capability requirements.

[0044] Through edge-cloud collaboration, the intelligent operation and maintenance big data center can acquire edge sensor data of corresponding vehicles on demand to optimize the intelligent model. This enables the optimization and customization of intelligent algorithm models for each vehicle and device based on actual operating conditions. The intelligent algorithm models are then distributed to the corresponding edge servers, which uniformly update the intelligent algorithm models of the edge controllers. This ensures that the intelligent algorithm models of each vehicle and device are consistent, which is highly beneficial for improving the availability and safety of vehicles and equipment, as well as enhancing the accuracy of corresponding fault warnings.

[0045] 4. Through end-to-end collaboration, the end server on the vehicle can connect and communicate with the end server of other vehicles within a set distance via its own wireless communication. It can intelligently sense the health status and fault information of vehicles within the set distance in real time, avoid potential safety hazards in a timely manner, and ensure the stable, safe and efficient operation of the vehicle. Attached Figure Description

[0046] Figure 1 This is a system block diagram of the present invention. Detailed Implementation

[0047] The specific embodiments of the present invention are described below to enable those skilled in the art to understand the present invention. However, it should be understood that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, various changes are obvious as long as they are within the spirit and scope of the present invention as defined and determined by the appended claims. All inventions utilizing the concept of the present invention are protected.

[0048] like Figure 1 As shown, the intelligent operation and maintenance system for rail transit based on edge computing and machine learning includes an end service layer, an edge service layer, and a cloud service layer.

[0049] The end-service layer is located on the vehicle and includes end sensors, end controllers, end servers, and the vehicle TCMS system;

[0050] The edge service layer is located in the ground control center and includes the intelligent line operation and maintenance system, the production business management system, and the intelligent vehicle maintenance system.

[0051] The cloud service layer includes an intelligent operation and maintenance big data center;

[0052] The end sensor is located at the position of the component to be monitored in the vehicle and is used to collect data from the component to be monitored;

[0053] The end controller connects the end sensor and the end server, and is connected to the vehicle TCMS system. It is used to acquire data of the monitored components collected by the end sensor, and to process the collected data in real time through an optimized machine learning algorithm model, and to generate corresponding information and control commands based on the processing results.

[0054] The end server connects the end controller to the intelligent operation and maintenance system of the line, and is connected to the vehicle TCMS system and the end servers of other vehicles within a set distance. It is used to receive the working instructions of the intelligent operation and maintenance system of the line, receive and store the results of the end controller's real-time processing of the collected data through an optimized machine learning algorithm model, generate status information and event information, and upload them to the intelligent operation and maintenance system of the line and the end servers of other vehicles within a set distance.

[0055] The vehicle TCMS system is connected to the end controller to receive corresponding information and control commands generated by the end controller based on the processing results, and to control the vehicle according to the corresponding information and control commands.

[0056] The vehicle TCMS system is connected to the end server to receive the results of the end controller's real-time processing of the collected data through an optimized machine learning algorithm model. Based on the processing results, the system generates corresponding information and control commands and controls the vehicle accordingly.

[0057] The intelligent operation and maintenance system is connected to the end server, the production business management system, the vehicle intelligent maintenance system, and the intelligent operation and maintenance big data center. It is used to receive relevant instructions from the intelligent operation and maintenance big data center, generate corresponding work instructions for the end server, the production business management system, and the vehicle intelligent maintenance system, and upload the status information and event information received from the end server, the production business management system, and the vehicle intelligent maintenance system to the intelligent operation and maintenance big data center.

[0058] The intelligent operation and maintenance big data center is used to aggregate information from vehicles, power supply, signaling, platform screen doors, and track catenary systems, as well as information generated by the production business management system and the vehicle intelligent maintenance system. Based on the diagnostic knowledge base and expert knowledge base, it establishes machine learning algorithm models, receives status and event information through the intelligent line operation and maintenance system, performs health assessments and fault predictions, and generates relevant instructions to be issued to the intelligent line operation and maintenance system. Based on the data received from the end sensors, it updates the diagnostic knowledge base and expert knowledge base through the established machine learning algorithm models, obtaining and issuing updated and optimized machine learning algorithm models.

[0059] Each vehicle includes an end server and at least one end controller, which is located around the component to be monitored. Each end controller monitors at least one end sensor.

[0060] The cloud service layer is deployed in locations including the control center, the owner's group company, and the OEM's operation and maintenance center.

[0061] The intelligent operation and maintenance big data center is connected to at least one line intelligent operation and maintenance system; a line intelligent operation and maintenance system is connected to at least one vehicle; each end server is installed in the corresponding vehicle and is directly connected to or cascaded with the end controller of the corresponding vehicle.

[0062] End sensors include accelerometers, strain sensors, fiber optic sensors, temperature sensors, composite sensors, industrial cameras, millimeter-wave radar, and lidar; components to be monitored include the running gear, pantograph, traction system, doors, air conditioning, and brakes.

[0063] The production business management system includes a vehicle status management module, a platform equipment management module, an operation daily plan management module, a construction and maintenance management module, a key equipment location management module, and a smart handheld terminal module.

[0064] The intelligent vehicle inspection system includes an intelligent inspection robot, a vehicle side running gear inspection module, a vehicle 360° vision inspection module, and a wheelset equivalent cone inspection module.

[0065] The connection methods between the end sensor and the end controller include wired or wireless connection; the connection methods between the end controller and the end server include wired or wireless connection; the connection methods between the end server and the intelligent operation and maintenance system of the line and the end servers of vehicles within a set distance include wireless connection; the connection methods between the vehicle TCMS system and the end controller or end server include communication cable connection; and the connection methods between the intelligent operation and maintenance system of the line and the intelligent operation and maintenance big data center include wired or wireless connection.

[0066] This intelligent operation and maintenance method for rail transit based on edge computing and machine learning includes the following steps:

[0067] S1. Collect data of the vehicle's monitored components using end sensors and upload it to the end controller;

[0068] S2. Based on the data received from the components to be monitored, the collected data is processed in real time using an optimized machine learning algorithm model, and corresponding information and control commands are generated based on the processing results.

[0069] S3. The end server receives and stores the results of the end controller's real-time processing of the collected data through an optimized machine learning algorithm model, generates status information and event information, and uploads them to the intelligent operation and maintenance system of the line and the end server of other vehicles within a set distance.

[0070] S4. Receive and control the vehicle according to the corresponding information and control instructions generated by the end controller through the vehicle TCMS system; or receive and control the vehicle according to the information and control instructions generated by the end server.

[0071] S5. Receive status and event information from the end server, production business management system and vehicle intelligent maintenance system through the line intelligent operation and maintenance system and upload them to the intelligent operation and maintenance big data center.

[0072] S6. By aggregating system information through the intelligent operation and maintenance big data center, an operation and maintenance knowledge base and an algorithm model library are established; the system information includes information on vehicles, power supply, signaling, platform screen doors, and track catenary, as well as information collected from the production business management system and the vehicle intelligent maintenance system; the operation and maintenance knowledge base includes a diagnostic knowledge base and an expert knowledge base; and the algorithm model library includes machine learning algorithm models.

[0073] S7. The intelligent operation and maintenance big data center performs health assessment and fault prediction on the received status and event information, and generates relevant instructions to be sent to the line intelligent operation and maintenance system; based on the received system data, the diagnostic knowledge base and expert knowledge base are updated through the established machine learning algorithm model, and the updated and optimized machine learning algorithm model is obtained and sent out.

[0074] S8. Receive relevant instructions from the intelligent operation and maintenance big data center through the intelligent operation and maintenance system and generate corresponding work instructions for the end server, production business management system and vehicle intelligent maintenance system.

[0075] S9. The end server, in accordance with the work instructions, cooperates with the end controller, the vehicle TCMS system, and the end servers of other vehicles within a set distance to perform subsequent operation and maintenance work.

[0076] This invention achieves:

[0077] Multi-disciplinary data acquisition involves the unified collection of multi-dimensional data from various specialized equipment and operational systems within the rail transit system. This includes multiple systems such as rolling stock, tracks, power supply, communication, signaling, platform screen doors, escalators and elevators, plumbing and electrical systems, and trackside equipment. While each system exists independently, they are also interconnected. End sensors are used to collect data from monitored components in real time and transmit it to the corresponding end controllers.

[0078] Data governance involves edge computing processing of data and the establishment of a unified data transmission protocol. The edge controller supports multiple acquisition protocols, storing massive amounts of heterogeneous, multi-source device data from various sensors, including digital, text, image, and audio data. Data fusion is performed, and the collected data is processed using machine learning algorithms deployed from the cloud service layer. The processing results are then stored on the edge server. Based on these results, relevant information and control commands are sent directly or via the edge server to the vehicle's TCMS system. Simultaneously, the edge server establishes a unified data transmission protocol to connect with the intelligent line operation and maintenance system, uploading the processing results.

[0079] Through edge-end collaboration, the intelligent operation and maintenance system for railway lines simultaneously acquires data from the production business management system and the vehicle intelligent maintenance system. It establishes an effective screening mechanism for core status data and fault data, including data cleaning, data integration, data transformation, and data reduction. The data is screened, analyzed, and valid information is retained while noisy data is removed. The data is then uploaded to the intelligent operation and maintenance big data platform using a unified transmission protocol.

[0080] The vehicle TCMS system establishes communication by coordinating end-to-end communication. The end server also connects wirelessly to the end servers of vehicles within a set distance (nearby vehicles), establishing vehicle-to-vehicle communication and enabling the system to obtain real-time status and fault information of vehicles within the set distance. The vehicle TCMS system can obtain relevant information and control commands from the end controllers or end servers, and control the vehicles based on these commands in conjunction with the vehicle motion control system.

[0081] Data mining involves establishing a unified big data mining platform. Multi-disciplinary data in a standardized format is stored in the data platform. Big data mining is then performed based on the stored target-specific data, and corresponding subsystems are established to meet the specific needs of each owner. These subsystems primarily include an equipment status early warning system, an equipment health management system, a machine learning algorithm model system, a maintenance management optimization system, a third-party expert maintenance knowledge base system, and a spare parts management system. Artificial intelligence and machine learning are used to establish corresponding status detection, fault early warning, and lifespan prediction algorithm model libraries for the target. When a fault is initially predicted, the causes of the fault are analyzed by combining multi-disciplinary data and fault correlation analysis.

[0082] Establish an algorithm model library and construct a machine learning algorithm model library. In the intelligent operations and maintenance big data center, multiple target algorithm model libraries are established using artificial intelligence and machine learning. Simultaneously, the platform can retrieve stored data from the end server as needed, optimize the corresponding machine learning algorithm models based on the data, and then connect to the corresponding line intelligent operations and maintenance system to send the optimized algorithm model to the controller through the end server. This ensures that the model within the end controller is updated in a timely manner, improving the accuracy of early warnings and predictions.

[0083] Establish an operation and maintenance knowledge base, including the basic principles and theories of different equipment systems and accumulated operation and maintenance experience, forming a knowledge graph. After the operation and maintenance knowledge graph is established, mining based on the operation and maintenance knowledge graph can improve the knowledge coverage of the expanded expert operation and maintenance knowledge base. Through the operation and maintenance knowledge base, expert solutions are provided for on-site operation and maintenance. Operation and maintenance decision-makers adjust maintenance plans and schemes through the maintenance management optimization system and coordinate the materials required for maintenance in a timely manner through the spare parts management system.

[0084] Data visualization: A data visualization platform is established, providing various data visualization models. The analysis results from the intelligent operations and maintenance big data center are displayed for all users to view. The visualization models are integrated with operations and maintenance data, establishing topology structures for different equipment faults and utilizing root cause analysis methods to achieve remote fault location. Management can understand the overall operation and maintenance situation and make rapid decisions. The execution layer can transmit on-site inspection operation videos to the intelligent operations and maintenance big data center from a first-person perspective using VR or AR glasses. The intelligent operations and maintenance big data center can communicate with the execution layer in real time through voice, images, and annotations, improving efficiency.

[0085] This invention is a smart operation and maintenance system for rail transit based on edge computing and machine learning. It addresses the problems of decentralized construction, numerous interfaces, data silos, data transmission and processing delays, inconsistent prediction algorithm models for all vehicles that cannot be updated in a timely manner, and the waste of existing resources caused by the need to build a new system for new functional requirements.

[0086] By fully leveraging IoT technology, unified data collection can be achieved on vehicles, dynamically expanding the information collected and hardware capabilities, and enabling deep data integration to avoid information silos.

[0087] By leveraging edge computing technology, the end controller of each vehicle can process the collected data in real time through machine learning algorithm models, and generate corresponding vehicle TCMS system information and control commands based on the processing results. This enables the vehicle to process the data collected by the end sensors 24 hours a day and provide timely feedback. The data can be processed locally, meeting the requirements of network bandwidth and latency.

[0088] By applying machine learning algorithms to achieve sustainable optimization of vehicle equipment and operational strategies, and through edge-cloud collaboration, the intelligent line maintenance system interacts with the production business management system and the intelligent vehicle inspection system to acquire relevant data. Centralized big data mining is performed on vehicle and production-related data, and various models such as health management, fault warning, and lifespan prediction are established using artificial intelligence and machine learning. This guides the optimization of existing maintenance plans, reduces over-maintenance and under-maintenance of equipment, and prevents equipment and vehicles from operating with defects. While improving vehicle system safety, it also provides data-driven decision support for maintenance personnel. Furthermore, by constructing a big data maintenance knowledge base, expert data support is provided for on-site maintenance, addressing the issue of insufficient skills among maintenance personnel and reducing personnel capability requirements.

[0089] Through edge-cloud collaboration, the intelligent operation and maintenance big data center can acquire edge sensor data of corresponding vehicles on demand to optimize the intelligent model. This enables the optimization and customization of intelligent algorithm models for each vehicle and device based on actual operating conditions. The intelligent algorithm models are then distributed to the corresponding edge servers, which uniformly update the intelligent algorithm models of the edge controllers. This ensures that the intelligent algorithm models of each vehicle and device are consistent, which is highly beneficial for improving the availability and safety of vehicles and equipment, as well as enhancing the accuracy of corresponding fault warnings.

[0090] Through end-to-end collaboration, the end server on the vehicle can connect and communicate with the end server of other vehicles within a set distance via its own wireless communication. This enables it to intelligently sense the health status and fault information of vehicles within the set distance in real time, promptly avoid potential safety hazards, and ensure the stable, safe, and efficient operation of the vehicles.

Claims

1. A smart operation and maintenance system for rail transit based on edge computing and machine learning, characterized in that: It includes the terminal service layer, the edge service layer, and the cloud service layer; The end service layer is located on the vehicle and includes end sensors, end controllers, end servers, and the vehicle TCMS system; The edge service layer is located in the ground control center and includes the intelligent line operation and maintenance system, the production business management system, and the intelligent vehicle maintenance system. The cloud service layer includes an intelligent operation and maintenance big data center; The end sensor is located at the position of the vehicle component to be monitored and is used to collect data from the component to be monitored. The end controller connects the end sensor and the end server, and is connected to the vehicle TCMS system. It is used to acquire data of the monitored components collected by the end sensor, and to process the collected data in real time through an optimized machine learning algorithm model, and to generate corresponding information and control commands based on the processing results. The end server connects the end controller to the intelligent operation and maintenance system of the line, and is connected to the vehicle TCMS system and the end servers of other vehicles within a set distance. It is used to receive the working instructions of the intelligent operation and maintenance system of the line, receive and store the results of the end controller's real-time processing of the collected data through an optimized machine learning algorithm model, generate status information and event information, and upload them to the intelligent operation and maintenance system of the line and the end servers of other vehicles within a set distance. The vehicle TCMS system is connected to the end controller and is used to receive corresponding information and control commands generated by the end controller based on the processing results, and control the vehicle according to the corresponding information and control commands. The vehicle TCMS system is connected to the end server and is used to receive the results of the end controller processing the collected data in real time through an optimized machine learning algorithm model. Based on the processing results, the system generates corresponding information and control commands and controls the vehicle accordingly. The intelligent operation and maintenance system for the line is connected to the end server, the production business management system, the vehicle intelligent maintenance system, and the intelligent operation and maintenance big data center. It is used to receive relevant instructions from the intelligent operation and maintenance big data center, generate corresponding work instructions for the end server, the production business management system, and the vehicle intelligent maintenance system, and upload the status information and event information received from the end server, the production business management system, and the vehicle intelligent maintenance system to the intelligent operation and maintenance big data center. The intelligent operation and maintenance big data center is used to collect information including vehicles, power supply, signaling, platform screen doors and track catenary, as well as information generated by the production business management system and vehicle intelligent maintenance system. It establishes machine learning algorithm models based on diagnostic knowledge base and expert knowledge base, receives status information and event information through the line intelligent operation and maintenance system, performs health assessment and fault prediction, and generates relevant instructions to be issued to the line intelligent operation and maintenance system. Based on the data received from the terminal sensors, the diagnostic knowledge base and expert knowledge base are updated through the established machine learning algorithm model, and the updated and optimized machine learning algorithm model is obtained and distributed.

2. The intelligent operation and maintenance system for rail transit based on edge computing and machine learning according to claim 1, characterized in that: Each vehicle includes an end server and at least one end controller, which is located around the component to be monitored. Each end controller monitors at least one end sensor.

3. The intelligent operation and maintenance system for rail transit based on edge computing and machine learning according to claim 1, characterized in that: The cloud service layer is deployed in locations including the control center, the owner group company, and the OEM's operation and maintenance center.

4. The intelligent operation and maintenance system for rail transit based on edge computing and machine learning according to claim 1, characterized in that: The intelligent operation and maintenance big data center is connected to at least one line intelligent operation and maintenance system; one line intelligent operation and maintenance system is connected to at least one vehicle; each end server is installed in the corresponding vehicle and is directly connected to or cascaded with the end controller of the corresponding vehicle.

5. The intelligent operation and maintenance system for rail transit based on edge computing and machine learning according to claim 1, characterized in that: The end sensors include accelerometers, strain sensors, fiber optic sensors, temperature sensors, composite sensors, industrial cameras, millimeter-wave radar, and lidar; the components to be monitored include the running gear, pantograph, traction system, doors, air conditioning, and brakes.

6. The intelligent operation and maintenance system for rail transit based on edge computing and machine learning according to claim 1, characterized in that: The production business management system includes a vehicle status management module, a platform equipment management module, an operation daily plan management module, a construction and maintenance management module, a key equipment positioning management module, and a smart handheld terminal module.

7. The intelligent operation and maintenance system for rail transit based on edge computing and machine learning according to claim 1, characterized in that: The intelligent vehicle maintenance system includes an intelligent maintenance robot, a vehicle side running gear detection module, a vehicle 360° visual inspection module, and a wheelset equivalent cone detection module.

8. The intelligent operation and maintenance system for rail transit based on edge computing and machine learning according to claim 1, characterized in that: The connection methods between the end sensor and the end controller include wired or wireless connection; the connection methods between the end controller and the end server include wired or wireless connection; the connection methods between the end server and the intelligent operation and maintenance system of the line and the end servers of vehicles within a set distance include wireless connection; the connection methods between the vehicle TCMS system and the end controller or end server include communication cable connection; and the connection methods between the intelligent operation and maintenance system of the line and the intelligent operation and maintenance big data center include wired or wireless connection.

9. A method for intelligent operation and maintenance of rail transit based on edge computing and machine learning, characterized in that, Includes the following steps: S1. Collect data of the vehicle's monitored components using end sensors and upload it to the end controller; S2. Based on the data received from the components to be monitored, the collected data is processed in real time using an optimized machine learning algorithm model, and corresponding information and control commands are generated based on the processing results. S3. The end server receives and stores the results of the end controller's real-time processing of the collected data through an optimized machine learning algorithm model, generates status information and event information, and uploads them to the intelligent operation and maintenance system of the line and the end server of other vehicles within a set distance. S4. Receive and control the vehicle according to the corresponding information and control instructions generated by the end controller through the vehicle TCMS system; or receive and control the vehicle according to the information and control instructions generated by the end server. S5. Receive status and event information from the end server, production business management system and vehicle intelligent maintenance system through the line intelligent operation and maintenance system and upload them to the intelligent operation and maintenance big data center. S6. By aggregating system information through the intelligent operation and maintenance big data center, an operation and maintenance knowledge base and an algorithm model library are established; the system information includes information on vehicles, power supply, signaling, platform screen doors, and track catenary, as well as information collected from the production business management system and the vehicle intelligent maintenance system; the operation and maintenance knowledge base includes a diagnostic knowledge base and an expert knowledge base; and the algorithm model library includes machine learning algorithm models. S7. The intelligent operation and maintenance big data center performs health assessment and fault prediction on the received status and event information, and generates relevant instructions to be sent to the line intelligent operation and maintenance system; based on the received system data, the diagnostic knowledge base and expert knowledge base are updated through the established machine learning algorithm model, and the updated and optimized machine learning algorithm model is obtained and sent out. S8. Receive relevant instructions from the intelligent operation and maintenance big data center through the intelligent operation and maintenance system and generate corresponding work instructions for the end server, production business management system and vehicle intelligent maintenance system. S9. The end server, in accordance with the work instructions, cooperates with the end controller, the vehicle TCMS system, and the end servers of other vehicles within a set distance to perform subsequent operation and maintenance work.

Images