A method and system for intelligent ballastless track node monitoring based on digital twin

Abstract

The present invention relates to the field of railway intelligent operation and maintenance technology, and provides an intelligent ballastless track node monitoring method and system based on digital twin, including: conducting all-weather testing on the ballastless track, and transmitting the monitoring data using a remote transmission system; performing preprocessing and establishing a database; constructing and training a neural network model of the ballastless track under the action of multiple physical fields; integrating a physical data model, a neural network model and a virtual data model, and establishing a digital twin of the physical space entity of the ballastless track in the virtual space through mapping and reconstruction; establishing a three-dimensional interactive interface, and visualizing the real-time strain field, temperature field and train information of the ballastless track passing through the ballastless track; evaluating and predicting the service status of the ballastless track and the running train, and issuing an alarm according to the response specification limit. The present invention can monitor and predict the ballastless track node and the train passing through the point, and ensure the safety of the ballastless track service and the high-speed train operation.

Description

Technical Field

[0001] The present invention relates to the technical field of railway intelligent operation and maintenance, and in particular to an intelligent ballastless track node monitoring method and system based on digital twins. Background Art

[0002] In recent years, many high-speed lines and urban rail transit lines have been completed and put into operation. With the increase of operation time and the increase of track service life, more and more vehicles and track system diseases have appeared. The condition monitoring and maintenance of vehicles and track systems is a major problem currently faced.

[0003] When high-speed railways pass through complex geological areas such as soft rocks, mining areas, karst caves, and frozen soil, the foundation under the ballastless track structure is prone to deformation. The ballastless track structure system is prone to overall deformation of the foundation and relative deformation between the track structure layers as the foundation deforms. In addition, some ballastless tracks are built in extremely cold areas or areas with large temperature differences between day and night, which produce additional loads such as temperature cycles and freeze-thaw cycles. Environmental factors affect the long-term service health of the ballastless track structure and have an adverse effect on driving comfort and safety. In order to monitor and predict the various adverse effects of the ballastless track structure under the influence of the environment, study the transmission law of the foundation deformation to the various layers of the ballastless track, and the transmission law of the displacement between the various layers of the ballastless track, it is urgent to accurately grasp the laws of the track changes with the environment and the long-term reciprocating load of the train in terms of quantity, but the detection and calculation methods of the above related quantities are not perfect at present.

[0004] Digital twins create virtual models of physical entities in a digital way, use data to simulate the behavior of physical entities in the real environment, and add or expand new capabilities for physical entities through virtual-real interactive feedback, data fusion analysis, and iterative decision optimization. However, looking at the entire life cycle, the current application of digital twins is mainly concentrated in post-production, such as product operation and health management, while in areas such as railway operation and maintenance, current research on digital twins is still in its infancy. Therefore, how to use digital twins to improve ballastless track node monitoring is an urgent problem to be solved. Summary of the invention

[0005] The content of the present invention is to provide an intelligent ballastless track node monitoring method and system based on digital twin, which can solve the problems of insufficient intelligent monitoring and intelligent operation and maintenance methods of ballastless track.

[0006] According to an intelligent ballastless track node monitoring method based on digital twinning of the present invention, the method comprises the following steps:

[0007] 1) Conduct all-weather testing through the high-speed railway ballastless track monitoring system to collect the dynamic and static responses of each layer structure of the ballastless track under the action of environmental loads, train loads, and foundation deformation loads, as well as climate and environmental parameters;

[0008] 2) Transmitting the collected data using the ballastless track monitoring data transmission system;

[0009] 3) Preprocess the collected signals and establish a database for training the neural network;

[0010] 4) Construct and train a neural network model of ballastless track under the action of multiple physical fields, and verify the consistency between the output data of the neural network model and the physical data of the ballastless track;

[0011] 5) Integrate the physical data model, neural network model and virtual data model, and establish the digital twin of the physical space entity of the ballastless track in the virtual space through mapping and reconstruction;

[0012] 6) Establish a three-dimensional interactive interface to visualize the ballastless track data cloud map and the running status of the running train;

[0013] 7) Evaluate and predict the service status of ballastless track and running trains, and issue alarms according to specifications.

[0014] Preferably, in step 1), the ballastless track monitoring system includes various physical sensors, specifically: a number of embedded sensors, surface-mounted sensors and roadside sensors;

[0015] The embedded sensors include embedded strain sensors and embedded temperature sensors, which are respectively embedded in the track slab or ballast slab, CA mortar or self-compacting concrete, and the inside of the base of the ballastless track;

[0016] Surface-mounted sensors include strain gauges, displacement sensors, temperature sensors, and acceleration sensors, which are fixed on the surface of rails and track plates;

[0017] Roadside sensors include infrared thermal imaging, ultrasonic, infrared displacement tester and DIC;

[0018] Physical sensors are deployed at each physical sensing point of the ballastless track. A physical sensing point is a sensing point where sensors can be installed. Various physical sensors are used to test the dynamic and static data changes of temperature, stress or strain, displacement, etc. of each layer structure of the ballastless track under the additional effects of temperature load, train load, and foundation deformation load.

[0019] Preferably, in step 2), the ballastless track monitoring data transmission system includes a plurality of armored conductors, a dynamic data acquisition switch, a static data acquisition switch, an instrument protection device, a signal collector, a data temporary storage station and a 5G transmission channel;

[0020] The armored wire is used to connect physical sensors. The dynamic data acquisition switch is a strain sensor, which is installed on the roadside at a certain distance from the physical sensor. When the train passes by, the switch is turned on to start high-frequency data acquisition, and the data acquisition is turned off after the train moves away from the intelligent rail node. The static acquisition switch is a timing electromagnetic relay, which collects data every certain period of time. The signal is collected by the signal collector and transferred to the temporary data storage station, and the data in the temporary data storage station is transmitted to the analysis and modeling system through the 5G transmission channel.

[0021] Preferably, in step 3), the data preprocessing method includes low-pass and high-pass filtering, EMD, EEMD, VMD, CEEMDAN time-frequency analysis method;

[0022] Various physical sensors are used to collect physical sensing data on the ballastless track in real time to form a physical sensing data set, which is stored in a physical data storage unit in a database; sensing points where sensors cannot be installed are called virtual sensing points, and a mapping point is set for each virtual sensing point. The mapping point obtains mapping data on the virtual sensing point by constructing finite element and multi-body dynamics simulation to form a mapping data set, which is stored in a virtual data storage unit in a database.

[0023] Preferably, in step 4), the neural network model structure includes four layers: the first network layer is the input layer, the second network layer is the timing network layer, the third network layer is the fully connected layer, and the fourth network layer is the output layer; the neural network model is used to construct a mapping relationship between the physical data model and the virtual data model; and the neural network model is trained and optimized using an optimization algorithm, with the mean square error as the objective function of the model, and the mean square error is optimized to the minimum value to obtain the optimal neural network model.

[0024] Preferably, in step 4), a neural network model of ballastless track under the action of multiple physical fields is constructed and trained using the database described in step 3, and the training data required for constructing the neural network model is a historical data set of processing parameters at N physical sensing points stored in a historical data storage unit in the database and a historical data set consisting of processing parameters at the same position points of M virtual sensing points corresponding to the actual manufacturing system; a mapping model for mapping the actual ballastless track is generated based on the physical sensing data set and the mapping sensing data set;

[0025] The consistency between the output data of the ballastless track neural network model and the physical model data of the ballastless track is verified by verifying the sensor and the system identification unit. The verification sensor does not participate in the construction of the neural network model throughout the process and is only used for model verification. The system identification unit is used to identify the model accuracy of the mapping relationship constructed by the constructed neural network model, and use the input and output data of the neural network model for offline identification. The output data at the same virtual sensing point at the installation position of the verification sensor is compared with the error of the verification sensor to determine whether the output data of the neural network model and the physical entity of the ballastless track are equivalent, and the correction and optimization of the neural network model are completed.

[0026] Preferably, in step 5), real-time interaction and synchronous feedback between the digital twin and the physical entity are realized through a database interface based on a distributed architecture; based on multi-source heterogeneous data fusion and neural network algorithms, data cleaning and conversion of information flow, control flow, data flow and decision flow between physical space and virtual space are realized, and shared aggregation, fusion and iterative optimization are realized.

[0027] Preferably, in step 6), 3D modeling software and rendering tools are used to render, map and optimize the edges of the physical entity model; a visual cloud map containing ballastless track geometry information, dynamic and static physical fields, geometric information of the moving train, and various dynamic index curves is established.

[0028] Preferably, in step 7), the service status of the ballastless track and the running train is evaluated according to the specifications, including: predicting the development trend of unevenness and TQI changes, whether the maximum temperature of the ballastless track temperature gradient exceeds the limit, whether the maximum tensile stress exceeds the limit; whether the wheel-rail force and wheel load reduction rate exceed the limit, and whether the vehicle body acceleration meets the riding comfort requirements; if any value exceeds the specification limit, an alarm is issued.

[0029] The present invention provides an intelligent ballastless track node monitoring system based on digital twin, which adopts the above-mentioned intelligent ballastless track node monitoring method based on digital twin.

[0030] The beneficial effects of the present invention are:

[0031] The node monitoring method of the present invention can monitor and evaluate the basic information of the running train, dynamic axle weight, and wheel-rail force in real time; conduct long-term monitoring of the inter-layer relationship, stress state, and temperature distribution state of the ballastless track, and predict the degradation law of the service performance of the ballastless track; provide feedback on the status of the vehicle-track coupling system, and remind the staff to perform maintenance to achieve the effect of intelligent operation and maintenance. Based on the existing ballastless track, the present invention proposes an intelligent ballastless track node monitoring method in the form of a combination of buried, surface, and non-contact sensors and ballastless track, which monitors and collects dynamic and static data of the ballastless track for a long time, and conducts system analysis. The method used can quickly discover abnormal conditions of the ballastless track and running trains, and make a certain degree of prediction based on the data model. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 It is a flow chart of a method for monitoring intelligent ballastless track nodes based on digital twin in an embodiment;

[0033] Figure 2 A schematic top view of the arrangement of physical sensors in an embodiment;

[0034] Figure 3 It is a front view schematic diagram of the arrangement of physical sensors in the embodiment;

[0035] Figure 4 It is a left view schematic diagram of the arrangement of physical sensors in the embodiment;

[0036] Figure 5 Schematic diagram of data collection method in the embodiment;

[0037] Figure 6 Schematic diagram of the interaction between the physical entity and the digital twin in the embodiment. DETAILED DESCRIPTION

[0038] In order to further understand the content of the present invention, the present invention is described in detail in conjunction with the accompanying drawings and embodiments. It should be understood that the embodiments are only for explaining the present invention and are not intended to limit it.

[0039] Example

[0040] like Figure 1 As shown, this embodiment provides an intelligent ballastless track node monitoring method based on digital twin, which includes the following steps:

[0041] 1) Conduct all-weather testing through the high-speed railway ballastless track monitoring system to collect the dynamic and static responses of each layer structure of the ballastless track under the action of environmental loads, train loads, and foundation deformation loads, as well as climate and environmental parameters;

[0042] In step 1), the ballastless track monitoring system includes various physical sensors, specifically: a number of embedded sensors, surface-mounted sensors and roadside sensors; the sensor layout principle is to cover all layers of the ballastless track structure and key functional positions.

[0043] The embedded sensors include embedded strain sensors and embedded temperature sensors, which are respectively embedded in the track slab or ballast slab, CA mortar or self-compacting concrete, and the inside of the base of the ballastless track;

[0044] Surface-mounted sensors include strain gauges, displacement sensors, temperature sensors, and acceleration sensors, which are fixed on the surface of rails and track plates;

[0045] Roadside sensors include infrared thermal imaging, ultrasonic, infrared displacement tester and DIC;

[0046] Physical sensors are deployed at each physical sensing point of the ballastless track. A physical sensing point is a sensing point where sensors can be installed. Various physical sensors are used to test the dynamic and static data changes of temperature, stress or strain, displacement, etc. of each layer structure of the ballastless track under the additional effects of temperature load, train load, and foundation deformation load.

[0047] Various physical sensors can well meet the needs of long-term monitoring, conduct all-weather real-time monitoring and feedback. In addition, sensors are evenly installed on each layer of the ballastless track structure, which is conducive to evaluating the relationship between layers and better assisting the establishment of digital twin models and providing training data for neural network models.

[0048] 2) Transmitting the collected data using the ballastless track monitoring data transmission system;

[0049] In step 2), the ballastless track monitoring data transmission system includes a number of armored conductors, a dynamic data acquisition switch, a static data acquisition switch, an instrument protection device, a signal collector, a data temporary storage station and a 5G transmission channel;

[0050] The armored conductor is used to connect the physical sensor, and is buried in the concrete structure together with the physical sensor or fixed along the concrete surface of the track using a fixing tool.

[0051] The dynamic data acquisition switch is a strain sensor installed on the roadside at a certain distance from the physical sensor. When a train passes by, the switch is turned on to start high-frequency data acquisition, and the data acquisition is turned off after the train moves away from the smart track node.

[0052] The static acquisition switch is a timing electromagnetic relay that collects data once every certain period of time. The signal is collected by the signal collector and transferred to the data temporary storage station, and the data in the data temporary storage station is transmitted to the analysis and modeling system through the 5G transmission channel. The signal collector, data temporary storage station and 5G transmission channel and other equipment are installed outside the railway line guardrail for easy maintenance. The equipment is waterproof, shockproof and anti-destructive with instrument protection devices to ensure long-term and efficient data transmission.

[0053] Since collecting train signals requires high-frequency collection, if the collection is carried out uninterruptedly throughout the day, the amount of data transmission will be huge. This transmission system optimizes the data transmission efficiency, so that the signal collector only collects and transmits data when the train passes, ensuring the collection and efficient transmission of key data. The armored wire and 5G transmission channel ensure the long-term and real-time nature of data collection.

[0054] like Figure 2 , Figure 3 , Figure 4 As shown, in this embodiment, the embedded strain sensor 13 is buried directly below the ballastless track support platform, and the embedded distributed optical fiber sensor 12 is buried longitudinally and transversely inside the concrete structure such as the ballastless track slab, ballast bed slab, CA mortar, self-compacting concrete, and base; the embedded temperature sensor 11 is evenly and discretely buried in the ballastless track slab, ballast bed slab, CA mortar, self-compacting concrete, and base. It is used to sense the temperature change and dynamic and static strain change of the ballastless track under the action of temperature load, train load, and foundation deformation additional load of each layer structure of the ballastless track.

[0055] like Figure 5 As shown, the dynamic data acquisition switch 21 is buried in the track plate at a certain distance from the physical sensor 1. When the dynamic data acquisition switch 21 senses that the train 5 is passing, it transmits a signal to the signal collector 23 to start the acquisition work. When the train passes for a period of time, the signal collector 23 is turned off for acquisition. The collected signals are stored in the data temporary storage station 24 and are remotely transmitted through the 5G transmission channel 25. Figure 5 In the figure, 22 is an armored wire for transmitting the test data of the sensor, and 3 is a data receiving terminal for receiving the data of the temporary data storage station 24.

[0056] 3) Preprocess the collected signals and establish a database for training the neural network;

[0057] In step 3), the data preprocessing method includes low-pass and high-pass filtering, EMD, EEMD, VMD, and CEEMDAN time-frequency analysis methods; the signal collected in this embodiment is a train-track high-frequency impact signal, so the sampling frequency is too large. In order to eliminate the interference of high-frequency noise signals, data preprocessing is used to eliminate noise signals.

[0058] Various physical sensors are used to collect physical sensing data on the ballastless track in real time to form a physical sensing data set, which is stored in a physical data storage unit in a database; sensing points where sensors cannot be installed are called virtual sensing points, and a mapping point is set for each virtual sensing point. The mapping point obtains mapping data on the virtual sensing point by constructing finite element and multi-body dynamics simulation to form a mapping data set, which is stored in a virtual data storage unit in a database.

[0059] 4) Construct and train a neural network model of ballastless track under the action of multiple physical fields, and verify the consistency between the output data of the neural network model and the physical data of the ballastless track;

[0060] In step 4), the neural network model structure includes four layers: the first network layer is the input layer, the second network layer is the time series network layer, the third network layer is the fully connected layer, and the fourth network layer is the output layer; the neural network model is used to construct the mapping relationship between the physical data model and the virtual data model; and the optimization algorithm is used to train and optimize the neural network model, using the mean square error as the objective function of the model, and optimizing the mean square error to the minimum value to obtain the optimal neural network model.

[0061] In step 4), the database described in step 3 is used to construct and train a neural network model of ballastless track under the action of multiple physical fields. The training data required for constructing the neural network model is a historical data set of processing parameters at N physical sensing points stored in a historical data storage unit in the database and a historical data set composed of processing parameters at the same position points of M virtual sensing points corresponding to the actual manufacturing system; a mapping model for mapping the actual ballastless track is generated based on the physical sensing data set and the mapping sensing data set.

[0062] The consistency between the output data of the ballastless track neural network model and the physical model data of the ballastless track is verified by verifying the sensor and the system identification unit. The verification sensor does not participate in the construction of the neural network model throughout the process and is only used for model verification. The system identification unit is used to identify the model accuracy of the mapping relationship constructed by the constructed neural network model, and use the input and output data of the neural network model for offline identification. The output data at the same virtual sensing point at the installation position of the verification sensor is compared with the error of the verification sensor to determine whether the output data of the neural network model and the physical entity of the ballastless track are equivalent, and the correction and optimization of the neural network model are completed.

[0063] 5) Integrate the physical data model, neural network model and virtual data model, and establish the digital twin of the physical space entity of the ballastless track in the virtual space through mapping and reconstruction;

[0064] The physical data model is a model composed of the numerical values ​​collected by various sensors deployed on the actual ballastless track; the virtual data model is a virtual data model that is obtained through a large number of calculations and cannot be measured by physical sensors.

[0065] In step 5), real-time interaction and synchronous feedback between the digital twin and the physical entity are achieved through a database interface based on a distributed architecture; based on multi-source heterogeneous data fusion and neural network algorithms, data cleaning and conversion of information flow, control flow, data flow and decision flow between physical space and virtual space are achieved, and shared aggregation, fusion and iterative optimization are achieved.

[0066] By using digital twin models and neural network models, based on the data collected by physical sensors, new capabilities are added or expanded for physical entities through virtual-real interactive feedback, data fusion analysis, and decision-making iteration optimization. In the entire life cycle of ballastless track, it is used to build a mapping of the real-time status of ballastless track in physical space to digital space based on real-time status data, and realize the monitoring scenario of the real-time status of ballastless track and running trains.

[0067] 6) Establish a three-dimensional interactive interface to visualize the ballastless track data cloud map and the running status of the running train;

[0068] In step 6), 3D modeling software and rendering tools are used to render, map and optimize the edges of the physical entity model; a visual cloud map containing ballastless track geometry information, dynamic and static physical fields, geometric information of the running train, and various dynamic index curves is established. Figure 6 As shown, a simulation scene is built, the rendered physical entity model is imported into Unity 3D, the calculation results of the physics engine are used to drive the graphics rendering engine for rendering and drawing, a visual simulation model of the ballastless track is constructed, and the real-time calculated virtual data is integrated into a cloud map or a time curve chart to realize the visual modeling and data display of the digital twin.

[0069] 7) Evaluate and predict the service status of ballastless track and running trains, and issue alarms according to specifications;

[0070] In step 7), the service status of the ballastless track and the running train is evaluated according to the specifications, including: predicting the development trend of unevenness and TQI changes, whether the maximum temperature of the ballastless track temperature gradient exceeds the limit, whether the maximum tensile stress exceeds the limit; whether the wheel-rail force and wheel load reduction rate exceed the limit, and whether the vehicle body acceleration meets the riding comfort requirements; if any value exceeds the specification limit, an alarm is issued.

[0071] This embodiment provides an intelligent ballastless track node monitoring system based on digital twin, which adopts the above-mentioned intelligent ballastless track node monitoring method based on digital twin.

[0072] The intelligent ballastless track node monitoring method of this embodiment can monitor the train operation status in real time, including basic train operation information, wheel-rail force, fastener support reaction force, etc., and monitor the service status of ballastless track nodes for a long time, including interlayer relationship changes, displacement, stress strain, and temperature field. In addition, this embodiment can be used to establish a real-time update model based on digital twins through 5G remote and efficient transmission, and combined with a neural network model, it is used to further predict the service status of ballastless track nodes and vehicle driving status. The method used can quickly discover abnormal conditions on site.

[0073] The present invention and its embodiments are described schematically above, and the description is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. Therefore, if a person skilled in the art is inspired by it and designs a structural method and an embodiment similar to the technical solution without creativity without departing from the purpose of the invention, they shall all fall within the protection scope of the present invention.

Claims

1. An intelligent ballastless track node monitoring method based on digital twin, characterized by: The following steps are involved: 1) Conduct all-weather testing through the high-speed railway ballastless track monitoring system to collect the dynamic and static responses of each layer structure of the ballastless track under the action of environmental loads, train loads, and foundation deformation loads, as well as climate and environmental parameters; 2) Transmitting the collected data using the ballastless track monitoring data transmission system; 3) Preprocess the collected signals and establish a database for training the neural network; 4) Construct and train a neural network model of ballastless track under the action of multiple physical fields, and verify the consistency between the output data of the neural network model and the physical data of the ballastless track; 5) Integrate the physical data model, neural network model and virtual data model, and establish the digital twin of the physical space entity of the ballastless track in the virtual space through mapping and reconstruction; 6) Establish a three-dimensional interactive interface to visualize the ballastless track data cloud map and the running status of the running train; 7) Evaluate and predict the service status of ballastless track and running trains, and issue alarms according to specifications.

2. According to claim 1, a method for monitoring intelligent ballastless track nodes based on digital twins is characterized in that: In step 1), the ballastless track monitoring system includes various physical sensors, specifically: a number of embedded sensors, surface-mounted sensors and roadside sensors; The embedded sensors include embedded strain sensors and embedded temperature sensors, which are respectively embedded in the track slab or ballast slab, CA mortar or self-compacting concrete, and the inside of the base of the ballastless track; Surface-mounted sensors include strain gauges, displacement sensors, temperature sensors, and acceleration sensors, which are fixed on the surface of rails and track plates; Roadside sensors include infrared thermal imaging, ultrasonic, infrared displacement tester and DIC; Physical sensors are deployed at each physical sensing point of the ballastless track. A physical sensing point is a sensing point where a sensor can be installed. Various physical sensors are used to test the static data changes of temperature, stress or strain, and position movement of each layer structure of the ballastless track under the additional effects of temperature load, train load, and foundation deformation load.

3. According to claim 2, a method for monitoring intelligent ballastless track nodes based on digital twins is characterized in that: In step 2), the ballastless track monitoring data transmission system includes a number of armored conductors, a dynamic data acquisition switch, a static data acquisition switch, an instrument protection device, a signal collector, a data temporary storage station and a 5G transmission channel; The armored conductor is used to connect physical sensors. The dynamic data acquisition switch is a strain sensor, which is installed on the roadside at a certain distance from the physical sensor. When the train passes, the switch is turned on to start high-frequency data acquisition, and the data acquisition is turned off after the train moves away from the intelligent rail node. The static acquisition switch is a timing electromagnetic relay, which collects data every certain period of time. The signal is collected by the signal collector and transferred to the temporary data storage station. The data in the temporary data storage station is transmitted to the analysis and modeling system through the 5G transmission channel.

4. According to claim 3, a method for monitoring intelligent ballastless track nodes based on digital twins is characterized in that: In step 3), the data preprocessing methods include low-pass and high-pass filtering, EMD, EEMD, VMD, and CEEMDAN time-frequency analysis methods; Various physical sensors are used to collect physical sensing data on the ballastless track in real time to form a physical sensing data set, which is stored in a physical data storage unit in a database; sensing points where sensors cannot be installed are called virtual sensing points, and a mapping point is set for each virtual sensing point. The mapping point obtains mapping data on the virtual sensing point by constructing finite element and multi-body dynamics simulation to form a mapping data set, which is stored in a virtual data storage unit in a database.

5. According to claim 4, a method for monitoring intelligent ballastless track nodes based on digital twins is characterized in that: In step 4), the neural network model structure includes four layers: the first network layer is the input layer, the second network layer is the time series network layer, the third network layer is the fully connected layer, and the fourth network layer is the output layer; the neural network model is used to construct the mapping relationship between the physical data model and the virtual data model; and the optimization algorithm is used to train and optimize the neural network model, using the mean square error as the objective function of the model, and optimizing the mean square error to the minimum value to obtain the optimal neural network model.

6. The method for monitoring intelligent ballastless track nodes based on digital twin according to claim 5 is characterized in that: In step 4), a neural network model of ballastless track under the action of multiple physical fields is constructed and trained using the database described in step 3. The training data required for constructing the neural network model is a historical data set of processing parameters at N physical sensing points stored in a historical data storage unit in the database and a historical data set consisting of processing parameters at the same position points of M virtual sensing points corresponding to the actual manufacturing system; a mapping model for mapping the actual ballastless track is generated based on the physical sensing data set and the mapping sensing data set; The consistency between the output data of the ballastless track neural network model and the physical model data of the ballastless track is verified by verifying the sensor and the system identification unit. The verification sensor does not participate in the construction of the neural network model throughout the process and is only used for model verification. The system identification unit is used to identify the model accuracy of the mapping relationship constructed by the constructed neural network model, and use the input and output data of the neural network model for offline identification. The output data at the same virtual sensing point at the installation position of the verification sensor is compared with the error of the verification sensor to determine whether the output data of the neural network model and the physical entity of the ballastless track are equivalent, and the correction and optimization of the neural network model are completed.

7. The method for monitoring intelligent ballastless track nodes based on digital twin according to claim 6 is characterized in that: In step 5), real-time interaction and synchronous feedback between the digital twin and the physical entity are achieved through a database interface based on a distributed architecture; based on multi-source heterogeneous data fusion and neural network algorithms, data cleaning and conversion of information flow, control flow, data flow and decision flow between physical space and virtual space are achieved, and shared aggregation, fusion and iterative optimization are achieved.

8. The method for monitoring intelligent ballastless track nodes based on digital twin according to claim 7, characterized in that: In step 6), 3D modeling software and rendering tools are used to render, map and optimize the edges of the physical entity model; a visual cloud map containing ballastless track geometry information, dynamic and static physical fields, geometric information of the moving train, and various dynamic index curves is established.

9. The method for monitoring intelligent ballastless track nodes based on digital twin according to claim 8, characterized in that: In step 7), the service status of the ballastless track and the running train is evaluated according to the specifications, including: predicting the development trend of unevenness and TQI changes, whether the maximum temperature of the ballastless track temperature gradient exceeds the limit, whether the maximum tensile stress exceeds the limit; whether the wheel-rail force and wheel load reduction rate exceed the limit, and whether the vehicle body acceleration meets the riding comfort requirements; if any value exceeds the specification limit, an alarm is issued.

10. An intelligent ballastless track node monitoring system based on digital twin, characterized by: It adopts an intelligent ballastless track node monitoring method based on digital twin as described in any one of claims 1-9.

Images