A traction system intelligent monitoring method and system based on digital twin technology
By leveraging digital twin technology and cross-platform data interaction, a hardware-in-the-loop simulation platform for the traction system and a Twinbuilder multiphysics order reduction platform were built, solving the problems of limited monitoring range and insufficient real-time monitoring capabilities of rail transit traction systems, and achieving high-precision, real-time monitoring and analysis across the entire range.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- CRRC YONGJI ELECTRIC CO LTD
- Filing Date
- 2025-09-25
- Publication Date
- 2026-07-31
AI Technical Summary
In existing technologies, the monitoring range of rail transit traction systems is limited, data collection and analysis are difficult, and real-time monitoring capabilities are insufficient, making it difficult to achieve full-range real-time monitoring and efficient analysis.
Using digital twin technology, a hardware-in-the-loop simulation platform for the traction system and a Twinbuilder multiphysics field reduction platform are built. Cross-platform data interaction is achieved through a TCP communication module, and electrical, magnetic, and thermal parameters are calculated and output in real time. High-precision simulation is performed using FPGA and deep neural networks to achieve blind-spot-free coverage and dynamic adaptive monitoring.
It enables real-time monitoring across the entire range, improves the operational reliability of the traction system, solves the problems of insufficient monitoring range and real-time monitoring capabilities, reduces dependence on sensors, and improves monitoring efficiency and accuracy.
Smart Images
Figure CN121300159B_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of intelligent monitoring technology for traction systems, and specifically relates to an intelligent monitoring method and system for traction systems based on digital twin technology. Background Technology
[0002] With the rapid development and technological maturity of my country's rail transit traction system, its reliability assessment and intelligent monitoring technology have gradually attracted attention. As the core equipment and power source of urban rail transit vehicles, the traction system is characterized by variable operating environments, complex internal structures, and high equipment integration, accounting for 32.33% of all train failures. The operating environment of high-power urban rail traction systems is even more complex, making high-reliability monitoring of their operating status crucial for vehicle safety. Currently, various sensors (voltage, current, temperature, speed, pressure, etc.) are deployed in rail transit train traction systems, essential for controlling the closed-loop stable operation of the traction system. Considering the economic cost of traction system products and reducing the risk of instability caused by monitoring, no additional sensors or detection circuits are added to the system. This limits the effectiveness of single-point and quantitative monitoring, making it difficult to achieve full-range real-time monitoring. Therefore, designing a new, full-range intelligent real-time simulation and monitoring method for the traction system to improve the overall operational reliability of the traction system is essential.
[0003] The existing technology is as follows: Currently, as shown in the appendix Figure 1 The traction system shown includes voltage, current, temperature, and speed sensors. The data collected by these sensors is used to control the closed-loop stable operation of the traction system and monitor its operating status. The traction system generates a large amount of data during operation. This data, collected by the sensors, is stored by the traction system controller and then downloaded offline for manual processing and analysis to monitor the operating status of each component and part of the traction system during normal operation.
[0004] Disadvantages of existing technology: The disadvantages of this solution are: 1) Limited monitoring range: The coverage and functional types of monitoring equipment are limited by the number of sensors, deployment space, location, and external environment, making it impossible to perform full-range monitoring or obtain performance extreme values. 2) High difficulty in data collection and analysis: Existing monitoring systems typically generate a large amount of monitoring data, and manual data processing and analysis are time-consuming, tedious, and inefficient. 3) Insufficient real-time monitoring capability: Existing monitoring models rely heavily on manual observation, which may prevent the timely detection of anomalies and the making of decisions. Summary of the Invention
[0005] The purpose of this invention is to provide an intelligent monitoring method and system for traction systems based on digital twin technology, to address the technical problems in existing technologies such as: 1) limited monitoring range; 2) difficulty in data collection and analysis; and 3) insufficient real-time monitoring capabilities.
[0006] This invention is achieved using the following technical solution: A method for intelligent monitoring of a traction system based on digital twin technology includes the following steps: Step 1: Build a hardware-in-the-loop simulation platform for the traction system; Driven by actual data acquisition, the electrical signal parameters and dynamic parameters of the converter and motor are acquired in real time. Step 2: Build the Twinbuilder multiphysics reduction platform; Real-time calculation of electromagnetic losses of heating devices and motors in the converter cabinet, and temperature field parameters of the converter cabinet and motor; Step 3: Set up the TCP communication module; Real-time cross-platform data interaction between the traction system hardware-in-the-loop simulation platform and the Twinbuilder multiphysics reduction platform; The TCP communication module functions as follows: by building its own TCP communication module, it enables cross-platform real-time data interaction between the traction system semi-physical simulation platform (hardware in the loop) and the Twinbuilder multiphysics reduction platform (software simulation), breaking through the limitations of traditional single-platform simulation; and solving technical challenges such as high-frequency data transmission and cross-platform protocol adaptation.
[0007] Step 4: Establish an intelligent monitoring platform for the digital twin technology-driven system; Based on the interactive data of the TCP communication protocol, the system runs the traction system hardware-in-the-loop simulation platform and the Twinbuilder multiphysics reduction platform; it synchronously outputs real-time data and visualized monitoring results of the electrical, magnetic and thermal parameters of the traction system, realizing the intelligent monitoring function of the traction system.
[0008] The intelligent monitoring functions of the traction system include: 1. Utilizing digital twin technology, the traction system hardware-in-the-loop simulation platform and the Twinbuilder multiphysics reduction platform output real-time electrical and electromagnetic losses, temperature curves, and field distribution parameters of the converter / motor components, eliminating the reliance on sensors in traditional monitoring; 2. Achieving blind-spot-free coverage (covering all components and circuits), dynamic adaptability (no need to adjust sensor layout), and high-precision field monitoring (upgrading from single-point measurement to three-dimensional temperature and electromagnetic field visualization); 3. Real-time waveform output, replacing traditional manual processing methods and solving the pain points of time-consuming and error-prone manual analysis in existing technologies.
[0009] After the traction system hardware-in-the-loop simulation platform and the Twinbuilder multiphysics reduction platform were built, cross-platform data interaction was achieved by adding TCP clients and TCP servers to these platforms. Running both platforms simultaneously, the digital twin-based intelligent monitoring platform for the traction system can output real-time data on the electric, magnetic, and thermal multiphysics fields of the traction system, realizing intelligent monitoring of the traction system's status based on digital twin technology.
[0010] A traction system intelligent monitoring system based on digital twin technology, used to implement any of the above-mentioned intelligent monitoring methods for traction systems based on digital twin technology; This includes a traction system hardware-in-the-loop simulation platform, a Twinbuilder multiphysics order reduction platform, a TCP communication module, and a traction system intelligent monitoring platform based on digital twin technology. The traction system hardware-in-the-loop simulation platform mainly runs the hardware-in-the-loop model of the main circuit of the traction system and the hardware-in-the-loop model of train dynamics. The Twinbuilder multiphysics reduction platform mainly runs the multiphysics reduction model of the converter and the multiphysics reduction model of the motor. TCP Server and DataConnector modules are built in the traction system hardware-in-the-loop simulation platform and the Twinbuilder multiphysics reduction platform respectively to realize the interconnection and interoperability of data between the traction system hardware-in-the-loop simulation platform and the Twinbuilder multiphysics reduction platform.
[0011] This invention establishes a traction system hardware-in-the-loop simulation platform and a Twinbuilder multiphysics reduction platform, and develops a TCP communication module to realize real-time reception and transmission of data between the CPU unit network client port of the traction system hardware-in-the-loop simulator and the Twinbuilder model server port, thereby achieving cross-platform data interaction and ultimately realizing an intelligent monitoring method for traction systems based on digital twin technology.
[0012] In the intelligent monitoring platform of the traction system using digital twin technology, the traction semi-physical platform built using FPGA high-precision semi-physical simulation technology can output electrical signal parameters of any device and circuit in the converter and motor in real time. The Twinbuilder multi-physics reduction platform can output electromagnetic losses of the main heat-generating components of the converter, maximum, minimum and average temperature curves in the converter cabinet, temperature distribution cloud map in the converter cabinet, iron loss and copper loss of the motor stator and rotor, temperature curves of the motor stator, rotor and frame, maximum and average temperature curves inside the motor, and temperature distribution cloud map of the motor in real time.
[0013] This solution is a real-time simulation platform that is not affected by the number of sensors, their placement space, location, external environment, or limitations in coverage and function type, enabling real-time monitoring across the entire range. It can also perform real-time simulation analysis of key parameters such as electrical, magnetic, and thermal parameters of the entire traction system, upgrading the original single-point monitoring to field monitoring and solving the shortcomings of 1) and 3) in the existing technology. The solution can output waveform data of electrical, magnetic, and thermal parameters of the traction system in real time, solving the problem of time-consuming, cumbersome, and inefficient manual data processing and analysis required in technology 2). Attached Figure Description
[0014] Figure 1 A schematic diagram illustrating the existing technical solutions.
[0015] Figure 2 This is a flowchart illustrating the method of the present invention.
[0016] Figure 3 This diagram illustrates the hardware-in-the-loop simulation platform for the traction system of the present invention.
[0017] Figure 4 This represents the technical roadmap for the multiphysics order reduction model of the converter of this invention.
[0018] Figure 5 This represents the technical roadmap for the multiphysics-based reduced-order model of the motor of the present invention.
[0019] Figure 6 This diagram illustrates the TCP communication module of the present invention.
[0020] Figure 7 This diagram illustrates the block diagram of the intelligent monitoring platform for the traction system based on the digital twin simulation technology of this invention.
[0021] Figure 8 This is a comparison chart showing the speed of the train according to the present invention (accuracy of 99%).
[0022] Figure 9 This is a comparison chart showing the motor current (accuracy 95.93%) of the present invention.
[0023] Figure 10 This is a comparison chart showing the output torque (accuracy 94.68%) of the motor of the present invention.
[0024] Figure 11 This represents a three-dimensional diagram of the converter of the present invention.
[0025] Figure 12 This represents the reduced-order model of the converter loss field in this invention.
[0026] Figure 13 This represents the reduced-order model of the temperature field of the converter in this invention.
[0027] Figure 14This diagram illustrates the operation of the reduced-order model of the converter loss field and the reduced-order model of the converter temperature field in Twinbuilder.
[0028] Figure 15 This represents the reduced-order model of the motor loss field in this invention.
[0029] Figure 16 This represents the reduced-order model of the motor temperature field in this invention.
[0030] Figure 17 This diagram illustrates the reduced-order model of the motor loss field and the reduced-order model of the motor temperature field of the present invention running in Twinbuilder.
[0031] Figure 18 This invention describes the TCP RTD model and its settings.
[0032] Figure 19 This describes the DataConnector model and settings of the present invention.
[0033] Figure 20 This indicates the TCP communication settings between the hardware-in-the-loop simulation platform for the traction system of this invention and the Twinbuilder multiphysics reduction platform.
[0034] Figure 21 This refers to the intelligent monitoring platform for the traction system based on the digital twin technology of Embodiment 2 of the present invention.
[0035] Figure 22 This indicates the speed of the train in Embodiment 2 of the present invention (accuracy of 99%).
[0036] Figure 23 This indicates the motor current in Embodiment 2 of the present invention (accuracy of 94.6%).
[0037] Figure 24 This indicates the motor output torque (accuracy 94%) in Embodiment 2 of the present invention.
[0038] Figure 25 This indicates the intermediate capacitor voltage (accuracy 95%) in Embodiment 2 of the present invention.
[0039] Figure 26 This shows the current curves of the IGBT, reactor, copper busbar, and busbar in Embodiment 2 of the present invention.
[0040] Figure 27 The curves showing the change in stator and rotor resistance of the motor as a function of temperature are shown in Embodiment 2 of the present invention.
[0041] Figure 28 The diagram shows the loss curves of the IGBT, reactor, copper busbar, and busbar in Embodiment 2 of the present invention.
[0042] Figure 29The maximum, minimum, and average temperature curves of the converter in Embodiment 2 of the present invention are shown.
[0043] Figure 30 This diagram shows the internal temperature field distribution of the converter in Embodiment 2 of the present invention.
[0044] Figure 31 This diagram shows the internal temperature field distribution of the converter in Embodiment 2 of the present invention.
[0045] Figure 32 This shows the temperature curves of the stator coil, core, and frame in Embodiment 2 of the present invention.
[0046] Figure 33 The maximum and average temperature curves of the motor in Embodiment 2 of the present invention are shown.
[0047] Figure 34 This diagram illustrates the temperature distribution field of the motor in Embodiment 2 of the present invention. Detailed Implementation
[0048] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings. Example 1
[0049] A method for intelligent monitoring of a traction system based on digital twin technology includes the following steps: Step 1: Build a hardware-in-the-loop simulation platform for the traction system; using actual collected data as the driving force, collect electrical signal parameters and dynamic parameters of the converter and motor in real time.
[0050] The electrical signal parameters and dynamic parameters include: The voltage and current parameters of any device in the converter and the motor, including the intermediate capacitor voltage, motor current, motor electromagnetic torque and train speed, are compared with the actual acquired signals, and the error is less than 6%.
[0051] Step one is as follows: S11. Build a semi-physical model of the main circuit of the traction system and run the semi-physical model of the main circuit of the traction system on the FPGA board in simulator one. S12. Build a semi-physical model of train dynamics. Based on the train dynamics parameters and track parameters, build a semi-physical model of train dynamics. Run the built and debugged semi-physical model of train dynamics on the CPU board of simulator 2. S13. Run the semi-physical model of the traction system main circuit and the semi-physical model of train dynamics in their respective hardware-in-the-loop simulators, output electrical parameters in real time, and verify the correctness of the traction system hardware-in-the-loop simulation platform by comparing them with the actual collected data.
[0052] The traction system main circuit semi-physical model and the train dynamics semi-physical model interact with each other via a reflective memory communication protocol: the traction system main circuit semi-physical model outputs motor torque to the train dynamics semi-physical model, and the train dynamics semi-physical model, upon receiving the motor torque signal, feeds back the vehicle speed / rotational speed to the traction system main circuit semi-physical model for closed-loop control. High-precision FPGA semi-physical simulation technology is applied to build the traction system main circuit semi-physical model based on FPGA hardware in Matlab / Simulink. Nanosecond-level simulation accuracy is achieved through FPGA language, improving the dynamic response speed and computational accuracy of the traction system main circuit semi-physical model.
[0053] Traditional software simulation is limited by computational efficiency, while FPGA hardware acceleration technology breaks through the real-time bottleneck and provides a high-fidelity data source for the Twinbuilder multiphysics reduction platform.
[0054] In practical applications, the traction controller and the simulator are connected via hardwired wiring harnesses to exchange data such as analog (AO), digital (DIO), and PWM signals. The level, traction / braking conditions, and load parameters from the actual acquired data are used as inputs to the traction system hardware-in-the-loop (HILL) simulation platform for simulation experiments. The output parameters and waveforms of the HILL, such as train speed, motor current, and torque, are recorded and compared with the actual acquired data to verify the accuracy of the HILL.
[0055] Step 2: Build the Twinbuilder multiphysics order reduction platform; calculate in real time the electromagnetic losses of heating devices and motors in the converter cabinet, as well as the temperature field parameters of the converter cabinet and motor.
[0056] Step 2 specifically involves: S21, constructing a multi-physics field model for the motor; the multi-physics field model for the motor includes a model for reducing the motor loss field and a model for reducing the motor temperature field.
[0057] S211. Based on the three-dimensional model of the motor, build a full-order model of the electromagnetic loss field of the motor. Using the effective value of current, torque and speed as inputs, calculate the average value of motor loss cycle under multiple sets of different inputs, and output the stator and rotor iron loss and copper loss. S212. A reduced-order model of the motor loss field is generated by polynomial fitting. S213. Establish a full-order model of the motor temperature field, and use the stator and rotor iron losses and copper losses output from the reduced-order model of the motor loss field as input excitations for the full-order model of the motor temperature field. Simulate and output transient motor coil, iron core and frame temperature parameters. S214. A dynamic reduced-order model of the temperature of the coil, core, and frame, as well as the temperature field of the whole machine, is generated using a deep neural network. S215. The reduced-order model of the motor loss field and the reduced-order model of the motor temperature field are run in Twinbuilder software to form a multi-physics reduced-order model of the motor. At the same time, real-time data interaction is performed with the traction system hardware-in-the-loop simulation platform to form a field-circuit coupled real-time simulation.
[0058] S22. Construct a multi-physics field reduction model for the converter; the multi-physics field reduction model for the converter includes a reduction model for the converter loss field and a reduction model for the converter temperature field.
[0059] S221. Obtain the three-dimensional structure of the traction system converter and construct an accurate, physics-based full-order simulation model of the electromagnetic loss field of the converter. S222. The current parameters of the heating devices in the converter provide excitation for the heating devices in the full-order model of the electromagnetic loss field of the converter, output the electromagnetic loss of each heating device, and use the polynomial regression algorithm to output the reduced-order model of the electromagnetic loss field, providing various loss excitations for the temperature field of the converter. S223. Build a full-order model of the temperature field of the converter, with the input being the loss data obtained by reducing the order of the electromagnetic loss field of the converter. S224. Use neural network algorithms to reduce the full-order model of the converter temperature field to a one-dimensional physical prototype-level model. S225. The reduced-order model of the converter loss field and the reduced-order model of the converter temperature field are run in Twinbuilder to form a multi-field coupled model. Simultaneously, real-time data interaction is performed with the traction system hardware-in-the-loop simulation platform to form a real-time field-circuit coupled simulation. In specific implementation, the current parameters of the converter's heat-generating components can be output from the traction system hardware-in-the-loop simulation platform or obtained through actual data acquisition. Preferably, only the main heat-generating components—IGBT modules, copper busbars, busbars, and reactors—can be acquired.
[0060] S23. Run the field reduction model built above in Twinbuilder software; The electromagnetic loss field parameters are: the real-time calculated values of the stator and rotor iron losses, copper losses of the motor, and the electromagnetic losses of the heating devices in the converter cabinet. The temperature field parameters include: maximum, minimum, and average temperature curves within the motor / converter cabinet; temperature field distribution cloud map inside the motor and converter cabinet; and real-time calculated values of motor stator and rotor resistances as a function of temperature. A multiphysics-based reduced-order model of the motor is used to quickly calculate motor losses and temperature data during operation. Simultaneously, the temperature data is fed back to the semi-physical model of the traction system main circuit to achieve bidirectional coupling. The specific technical route is attached. Figure 5 As shown, the full-order simulation model of the converter's electromagnetic losses and heat is reduced to a reduced-order model that can be simulated in real time, ensuring accuracy while improving simulation efficiency. The specific technical approach is attached. Figure 4 As shown.
[0061] Based on Twinbuilder software, reduced-order models of motor loss field, motor temperature field, converter loss field, and converter temperature field are constructed to achieve joint real-time simulation of converter loss field and temperature field, and motor loss field and temperature field. The reduced-order algorithm significantly reduces the computational complexity of multiphysics simulation while maintaining high accuracy and meeting real-time requirements.
[0062] Step 3: Build the TCP communication module; realize cross-platform real-time data interaction between the traction system hardware-in-the-loop simulation platform and the Twinbuilder multiphysics reduction platform.
[0063] Step 3, setting up the TCP communication module, specifically involves: S31. In the host computer workstation of the traction system hardware-in-the-loop simulation platform, build a TCP Server module as a TCP client, set the server IP and network port to be communicated, and specify the total length and data type of the data to be sent and received. S32. Add the Dataconnector module as a TCP server to the host computer workstation running the Twinbuilder multiphysics reduction platform, and set the communication network port and the format for receiving and sending data; S33. Configure the IP addresses of the two platform's host computers to be on the same network segment, and ensure that the IP address and port number configured on the client side are consistent with those configured on the server side. When the client communicates with the server, it must execute the server-side data first, and then execute the client-side data. Once the handshake is successfully completed within the specified timeout period, data exchange can proceed, indicating successful communication.
[0064] The traction system hardware-in-the-loop simulation platform primarily runs the hardware-in-the-loop models of the traction system main circuit and train dynamics, while the Twinbuilder multiphysics reduction platform primarily runs the multiphysics reduction models of the converter and motor. To achieve data interconnection between the two platforms, a TCP communication module for the TCP network protocol between the traction system hardware-in-the-loop simulation platform and the Twinbuilder multiphysics reduction model needs to be built, as shown in the attached diagram. Figure 6 As shown.
[0065] The real-time data interaction includes: Data transmission between the client and server based on the TCP protocol; Supports synchronous updates of multiphysics models and hardware-in-the-loop data (timeout ≤ 1ms). The data exchange content includes multi-dimensional signal parameters such as electrical signals, electromagnetic losses, and temperature signals.
[0066] Step 4: Establish an intelligent monitoring platform for the digital twin technology-driven system; Based on the interactive data of the TCP communication protocol, the system runs the traction system hardware-in-the-loop simulation platform and the Twinbuilder multiphysics reduction platform; it synchronously outputs real-time data and visualized monitoring results of the electrical, magnetic and thermal parameters of the traction system, realizing the intelligent monitoring function of the traction system.
[0067] The specific implementation steps are as follows: S41, the Twinbuilder multiphysics reduction platform, as the server side of the TCP communication module, first runs the converter loss field reduction model, converter temperature field reduction model, motor loss field reduction model, and motor temperature field reduction model in the Twinbuilder multiphysics reduction platform. After the above models are initialized, it waits for the current of the heating device, motor speed, torque, and motor current parameters sent by the traction system hardware-in-the-loop simulation platform through the TCP communication module. At the same time, it outputs the motor stator and rotor temperatures to the traction system hardware-in-the-loop simulation platform in real time to realize the field-circuit coupling closed loop. S42. The traction system hardware-in-the-loop simulation platform, acting as a TCP client, needs to wait for the Twinbuilder multiphysics reduction platform to complete its initialization before running. It outputs the current of the heating device, the motor speed and torque, and the motor current parameters, and transmits them to the Twinbuilder multiphysics reduction platform through the TCP communication module. At the same time, it receives the motor stator and rotor temperatures transmitted from the Twinbuilder multiphysics reduction platform to achieve a closed loop of field-circuit coupling. In the S43 traction system hardware-in-the-loop simulation platform, the overhead line voltage, level, traction / braking conditions, and load conditions from 3600 seconds of data collected from an entire urban rail line are used as hardware-in-the-loop input conditions. The Twinbuilder multiphysics reduction platform is used to simultaneously conduct simulation tests on the traction system hardware-in-the-loop simulation platform. The traction system hardware-in-the-loop simulation platform outputs the converter electrical parameters and motor electrical parameters in real time. Simultaneously, due to the influence of the motor stator and rotor temperatures transmitted from the Twinbuilder multiphysics reduction platform, the motor stator and rotor resistance parameters in the traction system hardware-in-the-loop simulation platform also change with temperature. The Twinbuilder multiphysics reduction platform can output in real time: electromagnetic losses of the converter's heating components, maximum, minimum, and average temperature curves in the converter cabinet, and temperature distribution cloud map in the converter cabinet; iron losses and copper losses of the motor stator and rotor, average temperature curves of the motor stator, rotor, and frame, and motor temperature distribution cloud map.
[0068] As attached Figure 7As shown, the traction system hardware-in-the-loop simulation platform mainly runs the hardware-in-the-loop model of the traction system main circuit and the hardware-in-the-loop model of train dynamics. The Twinbuilder multiphysics reduction platform mainly runs the multiphysics reduction model of the converter and the multiphysics reduction model of the motor. TCP Server and DataConnector modules are built in the traction system hardware-in-the-loop simulation platform and the Twinbuilder multiphysics reduction platform respectively to realize the interconnection of data between the traction system hardware-in-the-loop simulation platform and the Twinbuilder multiphysics reduction platform. Example 2
[0069] The content and effects of this invention are specifically illustrated using a certain urban rail transit vehicle traction system and an actual line as an example.
[0070] Step 1: The traction system hardware-in-the-loop simulation platform is attached. Figure 3 As shown: (1) Building a semi-physical simulation model of the traction system: Based on the semi-physical model of the main circuit of the traction system of the subway model, a semi-physical model of the main circuit of the traction system was built in Matlab software. The semi-physical model of the main circuit of the traction system includes a DC circuit, an inverter, and a motor model. The semi-physical model of the main circuit of the traction system runs on the FPGA board in simulator 1. The FPGA language is used to build the semi-physical model of the main circuit of the traction system. The simulation accuracy can reach the nanosecond level, ensuring the system simulation accuracy.
[0071] (2) Building a semi-physical model of train dynamics: Based on the dynamic parameters of the subway train and the actual line parameters, a semi-physical model of train dynamics is built in Simulink software. The semi-physical model of train dynamics includes the vehicle dynamics model, the wheel-rail contact dynamics model, the train wind resistance model and the line model. The built and debugged semi-physical model of train dynamics is downloaded to the CPU board of the simulator and run.
[0072] (3) The traction system main circuit semi-physical model and the train dynamics semi-physical model interact with each other through the reflection memory communication protocol: the traction system main circuit semi-physical model outputs the motor torque to the train dynamics semi-physical model. After receiving the motor torque signal, the train dynamics semi-physical model performs simulation calculations and feeds back the vehicle speed / rotation to the traction system main circuit semi-physical model to achieve closed-loop control.
[0073] (4) The traction controller controls the semi-physical model of the main circuit of the traction system according to the actual working conditions of the train and network commands. The traction controller and the simulator are connected through hard wire harness to perform data interaction of analog, digital and PWM pulse signals.
[0074] (5) Use the level, traction / braking conditions and load parameters in the actual collected data as input conditions for the traction system hardware-in-the-loop simulation platform, and conduct simulation tests on the traction system hardware-in-the-loop simulation platform.
[0075] (6) Record the output data and waveforms of the traction system hardware-in-the-loop simulation platform, including train speed, motor current, and motor torque, and compare and analyze them with the actual collected data to verify the accuracy of the traction system hardware-in-the-loop simulation platform. Appendix Figure 8-10 This chart compares measured data with output parameters from a hardware-in-the-loop simulation. Accuracy calculations were performed on approximately 1200 sets of measured and hardware-in-the-loop data for three parameters: train speed, motor current, and motor torque. The accuracy calculation formula for each set of data is as follows:
[0076] Where Xsim_n and Xtest_n are the semi-physical simulation value and the measured data value of the nth group, respectively. Let n be the precision value of the nth data group, where n is the number of points compared, ranging from 1 to 1200; the average precision is calculated as follows. The average accuracy of the calculated train speed, motor current and motor torque were 99%, 95.9% and 94.68% respectively, which verified that the accuracy of the hardware-in-the-loop simulation platform was >94%.
[0077] Step 2: Twinbuilder Multiphysics Reduction Platform.
[0078] The multiphysics-based model for converters reduces the order of the full-order simulation model for electromagnetic losses, heat, etc., creating a reduced-order model that can be simulated in real time. This ensures accuracy while improving simulation efficiency. Specific operation steps are attached. Figure 4 As shown; (1) Obtain the three-dimensional structure of the traction system converter, attached Figure 11 A full-order simulation model of electromagnetic losses of a physics-based converter was constructed.
[0079] (2) A semi-physical model of the main circuit of the traction system was built in the semi-physical simulation platform of the traction system. The current parameters of the main heat-generating devices in the converter were output, providing excitation for the key heat-generating devices in the full-order electromagnetic loss model of the converter. The electromagnetic losses of each heat-generating device (IGBT module, copper busbar, busbar, reactor) were output, and the reduced-order electromagnetic loss model was output using a polynomial regression algorithm. The reduced-order model of the converter loss field can output the electromagnetic losses of the IGBT module, copper busbar, busbar and reactor in real time, providing electromagnetic loss excitation input for the converter temperature field. The reduced-order model of the converter loss field is attached. Figure 12 As shown.
[0080] (3) Build a full-order model of the temperature field of the converter, and input the loss data of the heating device obtained by reducing the order of the electromagnetic loss field of the converter.
[0081] (4) The full-order simulation model of the converter temperature field is reduced to a one-dimensional physical prototype-level model using a neural network algorithm and integrated into the Twinbuilder multiphysics reduction platform. This enables the observation of data or field states that are difficult to collect through sensors, and the real-time output of the converter's internal temperature and temperature field diagram. The reduced-order model of the converter temperature field is attached. Figure 13 As shown.
[0082] (5) As attached Figure 14 As shown, the reduced-order model of the converter loss field and the reduced-order model of the converter temperature field are run in Twinbuilder software and interact with the traction system hardware-in-the-loop simulation platform in real time to form a field-circuit coupling real-time simulation.
[0083] A multiphysics-based reduced-order model of the motor is used to quickly calculate the motor's losses and temperature rise during operation. Simultaneously, the temperature data is fed back to the semi-physical model of the traction system's main circuit to achieve bidirectional coupling. Specific steps are attached. Figure 5 As shown: (1) Based on the three-dimensional model of the motor, a full-order model of the electromagnetic loss of the motor is built. The effective value of current, torque and speed are used as inputs to calculate the average value of motor loss cycle under multiple different inputs and output the stator and rotor iron loss and copper loss.
[0084] (2) A reduced-order model of the motor loss field is generated by polynomial fitting, and the stator and rotor iron losses and copper losses are output. The reduced-order model of the motor loss field is shown in the attached figure. Figure 15 As shown.
[0085] (3) Establish a full-order model of the motor temperature field, and use the stator and rotor iron loss and copper loss output by the reduced-order model of the motor loss field as the input excitation of the full-order model of the flow field temperature field, and simulate and output the transient motor coil, iron core and frame temperature parameters. (4) Using ANSYS Twinbuilder's dynamic rom builder, generate dynamic reduced-order models of the temperature at key points such as the stator coil, core, and frame, as well as the overall temperature field. The reduced-order model of the motor temperature field is attached. Figure 16 As shown.
[0086] (5) As attached Figure 17 As shown, the reduced-order models of the motor loss field and the motor temperature field are run in Twinbuilder software to form a multiphysics reduced-order model of the motor. This model interacts with the traction system hardware-in-the-loop simulation platform in real time to form a field-circuit coupled real-time simulation.
[0087] Step 3: Setting up the TCP communication module The traction system hardware-in-the-loop simulation platform primarily runs the hardware-in-the-loop model of the traction system main circuit and the hardware-in-the-loop model of train dynamics. The Twinbuilder multiphysics reduction platform primarily runs the multiphysics reduction model of the converter and the multiphysics reduction model of the motor. To achieve remote data interconnection between the two platforms, a TCP network protocol communication system needs to be established. The specific process is detailed in the attached document. Figure 6 As shown: (1) The CPU network port of the traction system hardware-in-the-loop simulation platform is used as a remote client. A TCP Server module is built on the CPU host computer, and the IP address and network port of the server to be communicated are set. The total length and data type of the data to be sent and received are specified, as well as the timeout for waiting for the server response (set to 500us). The TCP Server RTD model and settings of the traction system hardware-in-the-loop simulation platform are attached. Figure 18 As shown.
[0088] (2) Add the Dataconnector module as the server for TCP network communication on the host computer workstation running the Twinbuilder multiphysics reduction platform, and set the communication network port and the format for receiving and sending data; the DataConnector model and settings of the Twinbuilder multiphysics reduction platform are attached. Figure 19 As shown.
[0089] (3) As attached Figure 20 As shown, after the TCP communication modules on both platforms are set up, the IP addresses of the host computers on both platforms must be on the same network segment, and the IP address and port number set on the client must be consistent with those set on the server. When the client communicates with the server, the server must be run first, followed by the client. Once the handshake is successful within the specified timeout period, data exchange can begin, indicating successful communication.
[0090] Step 4: Construction of an intelligent monitoring platform for the traction system based on digital twin technology; Combined with appendix Figure 3 and attached Figure 21 As shown, the traction system hardware-in-the-loop simulation platform mainly runs the hardware-in-the-loop model of the traction system main circuit and the hardware-in-the-loop model of train dynamics. The Twinbuilder multiphysics reduction platform mainly runs the multiphysics reduction model of the converter and the multiphysics reduction model of the motor. TCP Server and DataConnector modules are built in the traction system hardware-in-the-loop simulation platform and the Twinbuilder multiphysics reduction platform respectively to realize the interconnection of data between the traction system hardware-in-the-loop simulation platform and the Twinbuilder multiphysics reduction platform.
[0091] (1) The Twinbuilder multiphysics reduction platform, as the server side of the TCP communication module, should first run the converter loss field reduction model, converter temperature field reduction model, motor loss field reduction model, and motor temperature field reduction model in the Twinbuilder multiphysics reduction platform. After the above models are initialized, wait for the parameters such as the current of the converter heating device, motor speed, torque, and motor current sent by the traction system hardware-in-the-loop simulation platform through the TCP communication module. At the same time, output the motor stator and rotor temperatures to the traction system hardware-in-the-loop simulation platform to realize the field-circuit coupling closed loop.
[0092] (2) The traction system hardware-in-the-loop simulation platform acts as the client of the TCP communication module. It waits for the Twinbuilder multiphysics reduction platform to complete its initialization before running. It outputs parameters such as the current of the converter heating device, motor speed, torque and motor current, and transmits them to the Twinbuilder multiphysics reduction platform through the TCP communication module. At the same time, it receives the motor stator and rotor temperatures transmitted from the Twinbuilder multiphysics reduction platform to realize the closed loop of field-circuit coupling.
[0093] (3) In the traction system hardware-in-the-loop simulation platform, the overhead line voltage, level, traction / braking conditions and load conditions in the 3600 seconds of real data collected from the entire line of the urban rail vehicle are used as hardware-in-the-loop input conditions. The Twinbuilder multiphysics reduction platform is run and the traction system hardware-in-the-loop simulation platform is used for simulation tests.
[0094] The traction system hardware-in-the-loop simulation platform can output real-time waveforms of train speed, motor torque, motor current, capacitor voltage, and current waveforms of the main heat-generating components of the converter, as shown in the attached figure. Figure 22-26 As shown in the figure. These parameters, compared with the actual collected data, achieve an accuracy of 94%. Simultaneously, the traction system hardware-in-the-loop simulation platform outputs real-time converter electrical parameters (current of heating devices) and motor electrical parameters (motor speed and torque, motor current), which are sent to the Twinbuilder multiphysics reduction platform via a TCP communication module as input excitation for the reduced-order model. Furthermore, due to the influence of the motor stator and rotor temperatures transmitted from the Twinbuilder multiphysics reduction platform, the motor rotor resistance in the traction system hardware-in-the-loop simulation platform also changes with temperature. The waveforms of the motor parameters changing with temperature are shown in the attached figure. Figure 27 As shown in the waveform, the motor stator and rotor resistance parameters change in real time with the temperature of the motor stator and rotor.
[0095] The Twinbuilder multiphysics order reduction platform can output in real time: electromagnetic losses of the main heat-generating components of the converter, maximum, minimum, and average temperature curves in the converter cabinet, and temperature distribution cloud map in the converter cabinet. The data waveforms are shown in the attached diagrams. Figures 28-30As shown; the motor stator and rotor iron losses, copper losses, temperature curves of the motor stator, rotor, and frame, maximum and average internal temperature curves of the motor, temperature distribution cloud map of the motor, and data waveforms are attached in sequence. Figures 31-34 As shown.
[0096] The intelligent monitoring platform for traction systems based on digital twin technology can output in real time the electrical and electromagnetic losses, temperature curves, and temperature field data of key components of the traction system converter and motor, enabling multi-dimensional monitoring of the entire traction system's electrical, magnetic, and thermal aspects. Example 3
[0097] A traction system intelligent monitoring system based on digital twin technology is provided to implement the traction system intelligent monitoring method based on digital twin technology described in any of the above embodiments. This includes a traction system hardware-in-the-loop simulation platform, a Twinbuilder multiphysics order reduction platform, a TCP communication module, and a traction system intelligent monitoring platform based on digital twin technology. The traction system hardware-in-the-loop simulation platform mainly runs the hardware-in-the-loop model of the main circuit of the traction system and the hardware-in-the-loop model of train dynamics. The Twinbuilder multiphysics reduction platform mainly runs the multiphysics reduction model of the converter and the multiphysics reduction model of the motor. TCP Server and DataConnector modules are built in the traction system hardware-in-the-loop simulation platform and the Twinbuilder multiphysics reduction platform respectively to realize the interconnection and interoperability of data between the traction system hardware-in-the-loop simulation platform and the Twinbuilder multiphysics reduction platform.
[0098] The above method is realized by real-time field-circuit coupling closed-loop simulation of the traction system semi-physical simulation platform and Twinbuilder multiphysics order reduction platform. This method not only eliminates the need to add additional sensors and monitoring circuits, avoiding excessive economic costs, but also does not interfere with the main circuit, reduces the risk of instability caused by monitoring, and can monitor multiple parameters simultaneously.
[0099] Meanwhile, this method retains the known system mathematical model (main circuit model and three-dimensional full-order model of motor, etc.) as the framework of the twin model (intelligent monitoring platform of traction system), and also introduces a data-driven method to update and correct the twin model. This not only solves the problems of error and inflexibility of traditional mathematical models, but also avoids the multi-parameter coupling and excessively long iteration time caused by pure data-driven methods. In addition, the amount of data required is greatly reduced, which lowers the requirements for data acquisition rate and quantity.
[0100] This method enables real-time simulation analysis of key parameters such as electrical, magnetic, and thermal parameters of the entire traction system, upgrading the original single-point monitoring to field monitoring. It effectively compensates for the deficiencies of monitoring equipment, provides the traction system with a brand-new, full-range intelligent real-time simulation and monitoring method, and improves the operational reliability of the entire traction system.
[0101] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement the present invention. Although detailed descriptions have been provided with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments, and they should all be covered within the protection scope of the claims.
Claims
1. A traction system intelligent monitoring method based on digital twin technology, characterized in that: Includes the following steps: Step 1: Build a hardware-in-the-loop simulation platform for the traction system; Driven by actual data acquisition, the electrical signal parameters and dynamic parameters of the converter and motor are acquired in real time. Step 2: Build the Twinbuilder multiphysics reduction platform; Real-time calculation of electromagnetic losses of heating devices and motors in the converter cabinet, and temperature field parameters of the converter cabinet and motor; Specifically: S21. Construct a multiphysics-based reduced-order model of the motor; The multiphysics model of motors includes the model of motor loss field reduction and the model of motor temperature field reduction. S211. Based on the three-dimensional model of the motor, build a full-order model of the electromagnetic loss field of the motor. Using the effective value of current, torque and speed as inputs, calculate the average value of motor loss cycle under multiple sets of different inputs, and output the stator and rotor iron loss and copper loss. S212. A reduced-order model of the motor loss field is generated by polynomial fitting. S213. Establish a full-order model of the motor temperature field, and use the stator and rotor iron losses and copper losses output from the reduced-order model of the motor loss field as input excitations for the full-order model of the motor temperature field. Simulate and output transient motor coil, iron core and frame temperature parameters. S214. A dynamic reduced-order model of the temperature of the coil, core, and frame, as well as the temperature field of the whole machine, is generated using a deep neural network. S215. The reduced-order model of the motor loss field and the reduced-order model of the motor temperature field are run in Twinbuilder software to form a multi-physics reduced-order model of the motor. At the same time, real-time data interaction is performed with the traction system hardware-in-the-loop simulation platform to form a field-circuit coupled real-time simulation. S22. Construct a multiphysics reduced-order model of the converter; The multiphysics reduction model of the converter includes the order reduction model of the converter loss field and the order reduction model of the converter temperature field. S221. Obtain the three-dimensional structure of the traction system converter and construct an accurate, physics-based full-order simulation model of the electromagnetic loss field of the converter. S222. The current parameters of the heating devices in the converter provide excitation for the heating devices in the full-order model of the electromagnetic loss field of the converter, output the electromagnetic loss of each heating device, and use the polynomial regression algorithm to output the reduced-order model of the electromagnetic loss field, providing various loss excitations for the temperature field of the converter. S223. Build a full-order model of the temperature field of the converter, with the input being the loss data obtained by reducing the order of the electromagnetic loss field of the converter. S224. Use neural network algorithms to reduce the full-order model of the converter temperature field to a one-dimensional physical prototype-level model. S225. Run the reduced-order model of the converter loss field and the reduced-order model of the converter temperature field in Twinbuilder to form a multi-field coupled model. At the same time, perform real-time data interaction with the traction system hardware-in-the-loop simulation platform to form a real-time simulation of field-circuit coupling. S23. Run the field reduction model built above in Twinbuilder software; The loss field parameters are: real-time calculated values of iron loss, copper loss of motor stator and rotor, and electromagnetic loss of heating devices in converter cabinet; The temperature field parameters are: maximum, minimum, and average temperature curves inside the motor / converter cabinet; temperature field distribution cloud map inside the motor and converter cabinet; and real-time calculated values of motor stator and rotor resistance as a function of temperature. Step 3: Set up the TCP communication module; Real-time cross-platform data interaction between the traction system hardware-in-the-loop simulation platform and the Twinbuilder multiphysics reduction platform; Step 4: Establish an intelligent monitoring platform for the digital twin technology-driven system; Based on the interactive data of the TCP communication protocol, the traction system hardware-in-the-loop simulation platform and the Twinbuilder multiphysics reduction platform are used. It can output real-time data and visualized monitoring results of the electrical, magnetic and thermal parameters of the traction system in real time, realizing the intelligent monitoring function of the traction system.
2. The intelligent monitoring method of the traction system based on the digital twin technology according to claim 1, characterized in that: Step one is as follows: S11. Construct a semi-physical model of the main circuit of the traction system; S12. Construct a semi-physical model of train dynamics; S13. Run the semi-physical model of the traction system main circuit and the semi-physical model of train dynamics in their respective hardware-in-the-loop simulators, output electrical parameters in real time, and verify the correctness of the traction system hardware-in-the-loop simulation platform by comparing them with the actual collected data.
3. The intelligent monitoring method of the traction system based on the digital twin technology according to claim 2, characterized in that: S11. Construct a semi-physical model of the main circuit of the traction system; The semi-physical model of the traction system main circuit is run on the FPGA board in simulator one; S12. Construct a semi-physical model of train dynamics; A semi-physical model of train dynamics is built based on the train dynamics parameters and track parameters. The built and debugged semi-physical model of train dynamics is then run on the CPU board of simulator 2. S13. The semi-physical model of the traction system main circuit and the semi-physical model of train dynamics exchange data through a reflective memory communication protocol: The traction system main circuit semi-physical model outputs the motor torque to the train dynamics semi-physical model. After receiving the motor torque signal, the train dynamics semi-physical model feeds back the vehicle speed / rotation speed to the traction system main circuit semi-physical model for closed-loop control.
4. The intelligent monitoring method of a traction system based on digital twin technology according to claim 1, characterized in that: The electrical signal parameters and dynamic parameters include: The voltage and current parameters of any device in the converter and the motor, including the intermediate capacitor voltage, motor current, motor electromagnetic torque, and train speed signal, are compared with the actual acquired signals, and the error is less than 6%.
5. The intelligent monitoring method for a traction system based on digital twin technology according to claim 1, characterized in that: Step 3, setting up the TCP communication module, specifically involves: S31. In the host computer workstation of the traction system hardware-in-the-loop simulation platform, build a TCP Server module as a TCP client, set the server IP and network port to be communicated, and specify the total length and data type of the data to be sent and received. S32. Add the Dataconnector module as a TCP server to the host computer workstation running the Twinbuilder multiphysics reduction platform, and set the communication network port and the format for receiving and sending data; S33. Configure the IP addresses of the two platform host computers to be in the same network segment, and the IP address and port number set on the client must be consistent with the IP address and port number set on the server. When the client communicates with the server, the server data must be executed first, and then the client data must be executed. When the handshake is successfully completed within the specified timeout period, the communication is successful.
6. The intelligent monitoring method for a traction system based on digital twin technology according to claim 5, characterized in that: The real-time data interaction includes: Data transmission between the client and server based on the TCP protocol; Supports synchronous updates of multiphysics models and hardware-in-the-loop data; The data exchange content includes multi-dimensional signal parameters such as electrical signals, electromagnetic losses, and temperature signals.
7. The intelligent monitoring method for a traction system based on digital twin technology according to claim 1, characterized in that: Step 4: Establish an intelligent monitoring platform for the digital twin technology-driven system. The specific implementation steps are as follows: S41, the Twinbuilder multiphysics reduction platform, acting as the server side of the TCP communication module, first runs the converter loss field reduction model, converter temperature field reduction model, motor loss field reduction model, and motor temperature field reduction model in the Twinbuilder multiphysics reduction platform. After the above models are initialized, it waits for the current of the heating device, motor speed, torque, and motor current parameters sent by the traction system hardware-in-the-loop simulation platform through the TCP communication module. At the same time, it outputs the motor stator and rotor temperatures to the traction system hardware-in-the-loop simulation platform in real time to realize the field-circuit coupling closed loop. S42. The traction system hardware-in-the-loop simulation platform, acting as a TCP client, needs to wait for the Twinbuilder multiphysics reduction platform to complete its initialization before running. It outputs the current of the heating device, the motor speed and torque, and the motor current parameters, and transmits them to the Twinbuilder multiphysics reduction platform through the TCP communication module. At the same time, it receives the motor stator and rotor temperatures transmitted from the Twinbuilder multiphysics reduction platform to achieve a closed loop of field-circuit coupling. In the S43 traction system hardware-in-the-loop simulation platform, the overhead line voltage, level, traction / braking conditions, and load conditions from 3600 seconds of data collected from an entire urban rail vehicle line are used as hardware-in-the-loop input conditions. The Twinbuilder multiphysics reduction platform is used to simultaneously conduct simulation tests on the traction system hardware-in-the-loop simulation platform. The traction system hardware-in-the-loop simulation platform outputs the converter electrical parameters and motor electrical parameters in real time. At the same time, due to the influence of the motor stator and rotor temperatures transmitted from the Twinbuilder multiphysics reduction platform, the motor stator and rotor resistance parameters in the traction system hardware-in-the-loop simulation platform also change with temperature. The Twinbuilder multiphysics reduction platform can output in real time: electromagnetic losses of the heat-generating components of the converter, maximum, minimum and average temperature curves in the converter cabinet, and temperature distribution cloud map in the converter cabinet; iron losses and copper losses of the motor stator and rotor, average temperature curves of the motor stator, rotor and frame, and temperature distribution cloud map of the motor.
8. A traction system intelligent monitoring system based on digital twin technology, characterized in that: This method is used to implement the intelligent monitoring method for a traction system based on digital twin technology as described in any one of claims 1-7. This includes a traction system hardware-in-the-loop simulation platform, a Twinbuilder multiphysics order reduction platform, a TCP communication module, and a traction system intelligent monitoring platform based on digital twin technology. The traction system hardware-in-the-loop simulation platform mainly runs the hardware-in-the-loop model of the main circuit of the traction system and the hardware-in-the-loop model of train dynamics. The Twinbuilder multiphysics reduction platform mainly runs the multiphysics reduction model of the converter and the multiphysics reduction model of the motor. The TCPServer module and the DataConnertor module are built in the traction system hardware-in-the-loop simulation platform and the Twinbuilder multiphysics reduction platform respectively to realize the interconnection and interoperability of data between the traction system hardware-in-the-loop simulation platform and the Twinbuilder multiphysics reduction platform.