Digital twinning method of multi-source heterogeneous data feedback mapping for track structure long-term service state evaluation

Abstract

The invention provides a digital twinning method of multi-source heterogeneous data feedback mapping for track structure long-term service state evaluation. The method comprises the following steps: burying fiber grating sensors, stress-strain sensors and the like in a track structure and a lining to obtain data, and performing ballastless track damage condition simulation and vehicle dynamics response analysis under the action of basic deformation by adopting finite element software and multi-body dynamics software to form a simulation calculation result library; establishing a three-dimensional virtual model of structures such as a long-section ballastless track and a lining on site, respectively carrying out result correspondence and result correction on a simulation result library and the three-dimensional model by adopting a feedback coupling mapping correction method, and carrying out coupling feedback mapping verification between different simulation calculation results and actually measured data; accurate reflection and real-time visual presentation of stress characteristics of a track structure under the influence of large deformation of a foundation are achieved, and meanwhile real-time prediction and reasonable evaluation are conducted on changes of the stress state of the track structure.

Description

Technical Field

[0001] The present invention relates to the technical field of rail structure service status assessment, and in particular to a digital twin method for multi-source heterogeneous data feedback mapping for long-term service status assessment of rail structures. Background Art

[0002] During the construction and operation of ballastless track structures in areas with large foundation deformation, due to the large rigidity of the ballastless track structure in the large foundation deformation section, the ballastless track will undergo structural deformation after being affected by the foundation deformation, which will eventually act as an uneven excitation on the vehicle itself traveling on the track structure, and have a great impact on the dynamic performance response of the vehicle. At the same time, large foundation deformation will damage or even destroy the track structure itself, seriously affecting the long-term service health of the track structure.

[0003] At present, the existing detection and monitoring methods for foundation deformation sections such as roadbed frost heave and settlement areas have problems such as long monitoring cycle, delayed feedback time, and inability to accurately reflect the changes in the state of the ballastless track structure and evaluate the service life of the track structure. There are many problems such as the inability to timely reflect the evolution of the service performance of the track structure under the influence of foundation deformation and the impact on vehicle dynamics, the inability to timely feedback for ballastless track maintenance and vehicle safety assessment, and the weak connection between simulation data and measured data. At the same time, the simulation results cannot be directly used for the full-process track structure service state assessment and prediction.

[0004] The existing detection and monitoring methods for frost heave and settlement areas of roadbed are unable to realize the visualization and feedback of detection and monitoring data, cannot intuitively display the measurement position and measurement data of each measuring point, cannot comprehensively and accurately describe and cannot predict and effectively display the long-term changes of large deformation of the foundation, cannot realize real-time feedback visualization and effective correction of data, and at the same time, it is impossible to realize real-time minute-level reciprocating feedback correction and optimization calculation and analysis of simulation results between simulation results and on-site monitoring data. Summary of the invention

[0005] An embodiment of the present invention provides a digital twin method for multi-source heterogeneous data feedback mapping for long-term service status assessment of a track structure, so as to effectively predict and assess changes in the stress state of the track structure.

[0006] In order to achieve the above object, the present invention adopts the following technical scheme.

[0007] A digital twin method for multi-source heterogeneous data feedback mapping for long-term service status assessment of rail structures, comprising:

[0008] The ballastless track structure is monitored on site by sensors to obtain the structural deformation and displacement data and stress, strain, crack and gap width data of the ballastless track structure, wherein the resulting deformation and displacement data are used as input data of the finite element model, and the stress, strain, crack and gap width data are used as model correction parameters and as monitoring data to establish a monitoring data model;

[0009] A vehicle-ballastless track-foundation large deformation coupling analysis model is constructed by using ABAQUS finite element software and multi-body dynamics software, and the basic deformation parameter data of the ballastless track structure and the parameter data of the track structure and the vehicle itself are input into the vehicle-ballastless track-foundation large deformation coupling analysis model, and the vehicle-ballastless track-foundation large deformation coupling analysis model outputs vehicle dynamic response data;

[0010] Establishing a finite element simulation model for track structure service status analysis, inputting structural deformation, displacement data and foundation deformation parameter data of the ballastless track structure into the track structure service status simulation analysis model, outputting track structure service status data under different foundation deformations, and determining the graded warning threshold for safe service of the track structure according to the track structure service status evaluation index;

[0011] The digital twin model includes a finite element model, a virtual model and an actual model. The finite element model in the digital twin model is calculated and corrected by a simulation correction method based on coupled feedback of simulation calculation results and analysis data, so as to achieve mutual correspondence between the virtual model and the actual model in the digital twin model.

[0012] The vehicle-track coupling dynamics model and the track structure service status assessment model are used as simulation models, and the vehicle dynamics response data output by the vehicle-ballastless track-foundation large deformation coupling analysis model and the track structure service status data under different foundation deformations output by the track structure service status analysis model are used as simulation result data;

[0013] Inputting the measured transverse and vertical displacement monitoring data of the rails, the measured surface stress data of the track structure and the displacement data of the track structure, the measured data of the gap and crack width of the track structure, and the measured data of vehicle dynamics and track structure service status into the digital twin model, combining the vehicle dynamics simulation data and the track structure service status simulation data under different foundation deformations, and using the digital twin model to output the corrected track structure service status data and vehicle dynamics response data;

[0014] Use deep learning to predict and expand the revised digital twin model data, and present model calculation results in minutes;

[0015] The deep learning method is used to predict and linearly interpolate the simulation calculation results of the model. The simulation calculation results are predicted and linearly interpolated using the LSTM neural network to achieve full process coverage and full space analysis of the foundation deformation results from the time domain perspective.

[0016] Use digital twin model data combined with deep learning methods to evaluate and predict the service status of track structures;

[0017] The input data of the track structure service status analysis and evaluation model are the modified vehicle dynamics simulation data and track structure service status data calculated by the digital twin model. Combined with the graded warning threshold calculated by simulation in step 3, the output data is the judgment data on the safe service status of the track structure, that is, the safe service status of the track structure is reasonably divided.

[0018] Preferably, the method further comprises:

[0019] Realize the miniaturization of digital twin model data, establish a three-dimensional digital model of the foundation large deformation section, establish a data display platform, and realize the real-time display of digital twin data.

[0020] Preferably, the vehicle-ballastless track-foundation large deformation coupling analysis model is constructed by using ABAQUS finite element software and multi-body dynamics software, and the basic deformation parameter data of the ballastless track structure and the parameter data of the track structure and the vehicle itself are input into the vehicle-ballastless track-foundation large deformation coupling analysis model, and the vehicle-ballastless track-foundation large deformation coupling analysis model outputs vehicle dynamic response data, including:

[0021] According to the stress, strain and deformation results of the ballastless track structure calculated by ABAQUS software, the finite element simulation model was modified using field measured data, and the data for the next step of deep learning calculation was generated in combination with the digital twin model. In terms of obtaining simulation data for the computational simulation database, the finite element software ABAQUS and the multi-body dynamics software SIMPACK and UM were used, and the calculation efficiency was comprehensively considered to establish a vehicle-ballastless track-foundation large deformation coupling analysis model;

[0022] In the vehicle-ballastless track-foundation large deformation coupling analysis model, the rails are established as beam units or solid units as track structure facilities that are in direct contact with the moving train; the fasteners are established using connector units to simulate the stiffness and damping of the fastener system; the track plate, self-compacting concrete and base plate are all simulated using solid units, while the tunnel lining is simulated using solid units and some dimensions are simplified. The external rock mass is treated with grounding springs or the infinite element method to consider the supporting effect of the geotechnical structure. At the same time, the contact relationship between the various layers of the structure and the stress conditions of the layered structure can be considered;

[0023] The boundary conditions and load loading boundaries of the large deformation coupling analysis model of the vehicle-ballastless track-foundation are determined. The structural deformation data and stress deformation data of the ballastless track structure collected by the sensor are used as the calculation initial conditions in the calculation simulation database. The form and basic parameters of the foundation deformation are comprehensively judged based on the data of each part, and the wavelength and wave depth of the deformation are fitted for the large deformation condition of the foundation. The change form of the fault, the basic deformation parameters of the fault and the fault angle of the line itself are analyzed for the fault offset, and then the forced displacement is applied by a three-dimensional nonlinear curve to simulate the large deformation of the foundation.

[0024] According to the basic parameters of the vehicle in the large deformation section, the vehicle's motion model is established, and the wheel-rail contact relationship is established according to the Hertz nonlinear contact theory. The calculation of the wheel-rail vertical and lateral forces and the vehicle body acceleration evaluation indicators is realized. The vehicle dynamics response is compared and verified using existing research reports, and then the vehicle dynamics performance response under different basic large deformation amplitudes, different action positions and different vehicle speeds is analyzed and judged.

[0025] Preferably, the establishment of the track structure service status analysis model includes:

[0026] According to the elastic-plastic damage parameters of the material, combined with indoor experimental data, an elastic-plastic damage model is established. The elastic-plastic damage model is used to analyze the evolution law of the performance of the structural materials of the ballastless track structure under long-term operation conditions, and the changes in the mechanical properties of the ballastless track structure itself under long-term operation conditions are analyzed. The structural damage parameters of the ballastless track structure under long-term operation conditions are assigned, and damage evaluation criteria are used for different positions of the ballastless track structure. The sub-model technology and the extended finite element method, the cohesive force model and the fracture mechanics analysis method are used to calculate the surface cracks of the track plate and the location of the structural separation.

[0027] The simulation calculation of the ballastless track structure subjected to large foundation deformation within the large foundation deformation range is carried out to form a damage result library of the ballastless track structure under different large foundation deformations. The damage conditions of the ballastless track structure under the influence of different large foundation deformation amplitudes are fitted using a deep learning method, and the results of different damage conditions of the ballastless track structure are fully covered and automatically generated.

[0028] Preferably, the method of using simulation correction based on coupled feedback of simulation calculation results and analysis data to calculate and correct the finite element model in the digital twin model to achieve mutual correspondence between the virtual model and the actual model in the digital twin model includes:

[0029] Under the analysis of limited boundary conditions, existing constitutive relations and contact relationship calculation assumptions, the track structure force is monitored based on the field measured data, and the foundation large deformation curve and amplitude at this time are calculated based on the mapping relationship. The foundation deformation curve and amplitude in the monitoring data of the foundation deformation monitoring sensor at this time are compared, and the simulation model itself is reasonably modified in combination with the data analysis results of other sensors;

[0030] According to the real-time feedback of the simulation results, the constitutive relationship of the finite element simulation analysis model is corrected and adjusted. The deep learning and multi-objective optimization theory are used to judge the reasons for the differences between the field measured data and the calculated simulation results. The influencing factors of the differences between the simulation results and the measured data are analyzed, and the model correction method is used to make real-time corrections to the constitutive relationship with the greatest influence, realizing the integrated use of simulation results and field measured results.

[0031] Preferably, the deep learning method is used to predict and linearly interpolate the simulation calculation results of the model, and the simulation calculation results are predicted and linearly interpolated using an LSTM neural network to achieve full process coverage and full space analysis of the basic deformation results from a time domain perspective, including:

[0032] A simulation result library establishment method that combines finite element fast simulation calculation technology and deep learning data analysis methods is developed. Fast calculation and simulation analysis are carried out on the established finite element model. A deep learning method is used to consider predicting and linearly interpolating the simulation calculation results of the model. The simulation results are predicted and linearly interpolated using an LSTM neural network. From the time domain perspective, full-process coverage and full-stage simulation analysis of the results of foundation deformation are achieved, thus realizing full-process structural service performance simulation.

[0033] By burying fiber grating sensors and stress-strain sensors in the track structure and lining to acquire data, finite element software and multi-body dynamics software are used to simulate the damage of ballastless track under foundation deformation and analyze the vehicle dynamic response. A simulation calculation result library is established, and a three-dimensional virtual model of the on-site long-section ballastless track and lining and other structures is established. The simulation result library and the three-dimensional model are respectively matched and corrected by the feedback coupling mapping correction method. The coupling feedback mapping verification is performed between different simulation calculation results and measured data, so as to achieve accurate reflection and real-time visualization of the stress characteristics of the track structure under the influence of large foundation deformation, and at the same time, the changes in the stress state of the track structure are predicted and reasonably evaluated in real time.

[0034] It can be seen from the technical solutions provided by the above-mentioned embodiments of the present invention that the method of the present invention can achieve accurate reflection and real-time visualization of the stress characteristics of the track structure under the influence of large deformation of the foundation, and at the same time make real-time prediction and reasonable evaluation of the changes in the stress state of the track structure.

[0035] Additional aspects and advantages of the present invention will be given in part in the following description, which will become obvious from the following description, or may be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0037] Figure 1 A schematic diagram of a digital twin method for multi-source heterogeneous data feedback mapping for long-term service status assessment of a ballastless track structure provided by an embodiment of the present invention;

[0038] Figure 2 A processing flow chart of a digital twin method for multi-source heterogeneous data feedback mapping for long-term service status evaluation of a ballastless track structure provided by an embodiment of the present invention;

[0039] Figure 3 A technical roadmap for a digital twin method for multivariate heterogeneous data feedback mapping provided by an embodiment of the present invention;

[0040] Figure 4 A schematic diagram of a method for establishing a simulation result library combining finite element fast simulation calculation technology and deep learning data analysis method provided by an embodiment of the present invention;

[0041] Figure 5 and Figure 6 A schematic diagram of wheel-rail vertical lateral force simulation analysis results in a large foundation deformation section provided by an embodiment of the present invention;

[0042] Figure 7 and Figure 8 A schematic diagram of simulation analysis results of vertical displacement and vertical acceleration of rails at different positions provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0043] The embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be interpreted as limiting the present invention.

[0044] It will be understood by those skilled in the art that, unless expressly stated, the singular forms "one", "said", and "the" used herein may also include plural forms. It should be further understood that the term "comprising" used in the specification of the present invention refers to the presence of the features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof. It should be understood that when we refer to an element as being "connected" or "coupled" to another element, it may be directly connected or coupled to the other element, or there may be intermediate elements. In addition, the "connection" or "coupling" used herein may include wireless connection or coupling. The term "and / or" used herein includes any unit and all combinations of one or more associated listed items.

[0045] It will be understood by those skilled in the art that, unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as those generally understood by those skilled in the art in the art to which the present invention belongs. It should also be understood that terms such as those defined in common dictionaries should be understood to have meanings consistent with the meanings in the context of the prior art, and will not be interpreted with idealized or overly formal meanings unless defined as herein.

[0046] To facilitate understanding of the embodiments of the present invention, several specific embodiments will be further explained below with reference to the accompanying drawings, and each embodiment does not constitute a limitation on the embodiments of the present invention.

[0047] The present invention proposes a digital twin method for multi-source heterogeneous data feedback mapping for long-term service status evaluation of track structures. The method adopts the concept of digital twin to realize service status monitoring, evaluation and effective visualization of track structures in sections with large foundation deformation.

[0048] The implementation principle of the digital twin method of multi-source heterogeneous data feedback mapping for long-term service status evaluation of ballastless track structure proposed in the embodiment of the present invention is as follows: Figure 1As shown, a vehicle-track coupling dynamics model considering large deformation of the foundation is established by using a finite element simulation analysis method, and simulation calculations considering large deformation of soft rock and active fault zones are performed respectively. The simulation results are first placed in a preliminary result library of the simulation results, and at the same time, field data are collected and stored, and differences in multi-source data types are unified. Then, the calculation results are first checked with reference to existing monitoring data, and then a digital twin method of multi-source heterogeneous data feedback mapping is used to correct the existing simulation results and the contact relationship, constitutive relationship, etc. of the simulation model, and the results of the simulation calculation are miniaturized. At the same time, a deep learning method is used to expand the simulation results and realize the simulation results of the entire process. The simulation results and the calculation results expanded by the deep learning method are imported into the simulation result library, and a three-dimensional visualization display platform is used to display the simulation results in real time.

[0049] The processing flow of a digital twin method for multi-source heterogeneous data feedback mapping for long-term service status evaluation of ballastless track structure proposed in an embodiment of the present invention is as follows: Figure 2 As shown, the processing steps include the following:

[0050] Step S1, monitoring the ballastless track structure on site through sensors to obtain structural deformation and displacement data and stress, strain, crack and gap width data of the ballastless track structure, wherein the resulting deformation and displacement data are used as input data of the finite element model, and the stress, strain, crack and gap width data are used as model correction parameters, and at the same time as monitoring data to establish a monitoring data model.

[0051] For the large deformation section of the ballastless track structure, a long-section fiber Bragg grating sensor is used to monitor the vertical and lateral deformation of the long-section structure. Stress sensors are buried in the ballastless track structure to monitor the changes in the longitudinal and lateral stresses of the structure. Video sensors are used to monitor the relative deformation of the ballastless track structure under vehicle operation. The macro deformation data of the long section of the track structure under the action of large foundation deformation of the ballastless track and the structural deformation data of the track plate in the short curved section are collected.

[0052] Step S2: construct a vehicle-ballastless track-foundation large deformation coupling analysis model using finite element software such as ABAQUS and multi-body dynamics software.

[0053] The input data of this step are the parameter data of the foundation deformation and the parameter data of the track structure and the vehicle itself. The foundation deformation acts as the track irregularity excitation of the moving train and affects the change of the vehicle dynamic response. The output data are the vehicle dynamic response parameters, including the vertical and lateral acceleration of the vehicle body, the vertical and lateral force of the wheel and rail, the displacement of the track structure and the vibration acceleration.

[0054] According to the stress, strain and structural deformation results of the ballastless track structure calculated by ABAQUS software, the finite element simulation model is corrected using the on-site measured data, and the digital twin model is combined to generate data for the next step of deep learning calculation. For example, the stress data of the track structure under the influence of different foundation deformation parameters is generated, which can be used for time series prediction. At the same time, the foundation deformation parameters and foundation deformation types are identified based on the differences in the vehicle dynamics calculation data and the track structure dynamic response data under normal operation and with the influence of foundation deformation.

[0055] (1) In terms of simulation data acquisition in the computational simulation database, the finite element software ABAQUS and the multi-body dynamics software SIMPACK and UM are used to comprehensively consider the computational efficiency and establish a refined vehicle-ballastless track-foundation large deformation coupling analysis model. In the vehicle-ballastless track-foundation large deformation coupling analysis model, the rails are established as beam units or solid units as track structure facilities that are in direct contact with the moving train; the fasteners are established as connector units, which can accurately simulate the stiffness and damping of the fastener system; the track plate, self-compacting concrete and base plate are all simulated using solid units, while the tunnel lining is simulated using solid units, and some dimensions are simplified. The external rock mass is used with grounding springs or infinite element method to consider the supporting role of the geotechnical structure, and the contact relationship between the various layers of the structure and the stress of the layered structure can be considered.

[0056] The vehicle-ballastless track-foundation large deformation coupling analysis model can effectively supplement the amount of data used for deep learning, fill in a sufficient number of deep learning parameters, and ultimately form a sufficient amount and different dimensions of vehicle operation stability and track structure safe service data through field data verification and the formation of a digital twin model, which can provide a reliable data source for the prediction and identification of track structure status using deep learning methods.

[0057] (2) Based on the existing research experience, the boundary conditions and load loading boundaries of the large deformation coupling analysis model of the vehicle-ballastless track-foundation are determined. The structural deformation data and stress deformation data of the ballastless track structure collected by the sensor are used as the initial conditions for the calculation in the simulation database. The form and basic parameters of the foundation deformation are comprehensively judged based on the data of each part. For the large deformation condition of the foundation such as large deformation of soft rock, the wavelength and depth of the deformation can be fitted; for the fault offset, the basic deformation parameters such as the change form of the fault, the offset angle of the fault and the line itself can be analyzed, and then the three-dimensional nonlinear curve is used to impose forced displacement to achieve accurate simulation of the large deformation of the foundation.

[0058] (3) Based on the basic parameters of the vehicle in the large deformation section, the vehicle motion model is established, and the accurate wheel-rail contact relationship is established based on the Hertz nonlinear contact theory. The calculation of evaluation indicators such as the wheel-rail vertical and lateral forces and the vehicle body acceleration can be realized. The vehicle dynamic response is compared and verified using existing research reports, and then the vehicle dynamic performance response under different basic large deformation amplitudes, different action positions and different vehicle speeds is analyzed and judged.

[0059] Step S3: Establishing a track structure service status analysis model.

[0060] The input data of this step are the basic deformation parameters and the monitoring data obtained in step S1, the main parameters of the track structure, and there is no need to use the data output from step S2. The output track structure damage and crack generation rules are used in step S9. According to the vehicle dynamics response data (output data from steps S2 and S4) and the influence rules of the track structure service status (output data from steps S3 and S4), the graded warning thresholds for the safe service of the track structure are analyzed and used as part of the input data for step S4.

[0061] (1) Based on the elastic-plastic damage parameters of the material and combined with indoor experimental data such as fatigue tests, a refined elastic-plastic damage model is established. The refined elastic-plastic damage model is used to analyze the evolution of the performance of the structural materials of the ballastless track structure under long-term operation conditions. The changes in the mechanical properties of the ballastless track structure itself under long-term operation conditions are analyzed, and the structural damage parameters of the ballastless track structure under long-term operation conditions are assigned. Damage evaluation criteria are adopted for different positions of the ballastless track structure. Sub-model technology, extended finite element method, cohesive force model and fracture mechanics analysis methods are used to perform refined calculations of the surface cracks of the track plate and the positions of structural separations.

[0062] (2) Use supercomputers to simulate and calculate the conditions under which the ballastless track structure is subjected to large foundation deformation within the large foundation deformation range, and form a damage result library for the ballastless track structure under different large foundation deformations. Use deep learning methods to fit the damage conditions of the ballastless track structure under the influence of different large foundation deformation amplitudes, and automatically generate results covering all damage conditions of the ballastless track structure. Considering the simulation deviation that cannot be completely matched between the existing simulation and measured data, it is impossible to achieve long-term stability and accuracy evaluation of the simulation calculation and performance prediction of the ballastless track structure.

[0063] Step S4. Calculate and correct the finite element model in the digital twin model by using a simulation correction method based on coupled feedback of simulation calculation results and analysis data, so as to achieve mutual correspondence between the virtual model and the actual model in the digital twin model.

[0064] The input data include the measured lateral and vertical displacement monitoring data of the rails, the measured surface stress data of the track structure and the displacement data of the track structure (step S1 data), the vehicle dynamics simulation data (step S2 data) and the service status data of the track structure under different foundation deformations (step S3 data). The output data include the corrected stress strain and displacement data of the track structure, the vehicle dynamics simulation data and the service status data of the track structure.

[0065] The simulation model in this step supplements the basic data acquisition of the virtual model. The existing monitoring data is limited and cannot fully realize the establishment of the virtual model. Therefore, a simulation model is needed to fill and improve the data of the virtual model.

[0066] The simulation model uses the vehicle-track coupling dynamics model and the track structure service status assessment model built in step S2 and step S3, and the simulation result data is the data exported in step S2 and step S3.

[0067] (1) In the case of limited boundary conditions, existing constitutive relations, contact relations and other calculation assumptions, due to the complexity of the display model and the randomness of the structural load, the measured data and the simulation results cannot correspond in actual use, and the complete correctness of the numerical simulation technology cannot be fully realized.

[0068] (2) When arranging points for on-site monitoring, it is necessary to design special measuring points for verifying simulation results. For example, the stress on the track structure can be monitored based on the actual measured data on site, and the foundation deformation curve and amplitude at this time can be inferred based on the mapping relationship. At this time, the foundation deformation curve and amplitude in the monitoring data of the foundation deformation monitoring sensor can be compared. At the same time, the simulation model itself can be reasonably corrected in combination with the data analysis results of other sensors.

[0069] (3) According to the real-time feedback of the simulation results, the constitutive relationship of the finite element simulation analysis model needs to be effectively corrected and adjusted. Deep learning and multi-objective optimization theory are used to judge the reasons for the differences between the field measured data and the calculated simulation results. Due to the scale effect between the indoor experimental model and the field structure, there may be problems with inaccurate calculation of the constitutive relationship. The main influencing factors for the differences between the simulation results and the measured data are the set contact relationship and other constitutive relationships. The model correction method is used to make real-time corrections to the constitutive relationship with the greatest impact, realizing the integrated use of simulation results and field measured results.

[0070] Step S5. Use simulation data to visualize the internal stress of the track structure and fully display the structural force.

[0071] Due to the limitations of sensor arrangement in the track structure, existing sensing technologies and data analysis methods are unable to achieve stress analysis and visualization inside the track structure, especially for some structures with complex structures and certain thickness. Therefore, it is necessary to use field monitoring data and simulation results to verify the accuracy of the simulation, so as to achieve accurate calculation of the stress inside the track structure, reflect the real-time stress inside the track structure that cannot be reliably detected, and simulate and predict the structural stress in hidden locations such as grooves inside some structures.

[0072] Step S6. Use deep learning to predict the results of the simulation analysis and realize the visualization of the minute-level model calculation results.

[0073] (1) The simulation result library establishment method combines finite element fast simulation calculation technology and deep learning data analysis method, mainly for the establishment of the completed finite element model to carry out fast calculation and simulation analysis. At the same time, due to the simulation calculation step and time limit of the established entity refined finite element model, only hour-level finite element simulation can be achieved, and the full process of simulation results cannot be covered in minutes, and the full process of structural stress and other calculation results cannot be visualized.

[0074] (2) A deep learning method is used to consider the reasonable prediction and linear interpolation of the simulation calculation results of the model. The simulation calculation results are predicted and linearly interpolated using an LSTM neural network. From the time domain perspective, full-process coverage and full-stage simulation analysis of the basic deformation results are achieved, which can be used for the full-process structural service performance simulation.

[0075] Step S7: Realize the miniaturization of simulation result data and the prediction of simulation results.

[0076] For large quantities of simulation data, classified storage and data format processing are implemented to achieve miniaturization of simulation data and rapid reading and use of calculation results. The data in the simulation model is edited and exported to the calculation result display software. In addition to data conversion of the results of different units themselves and the relationship between different units, the coordinate conversion and correction of data such as grid position are performed. The result file is compressed and format converted to form a result library that can realize frame-by-frame reading of simulation calculation results such as deformation and stress of ballastless track structure.

[0077] Step S8: Establish a three-dimensional digital model of the large deformation section of the foundation and establish a simulation result display platform.

[0078] The 3D simulation results are visualized in combination with 3D visualization software. At the same time, 3D point cloud technology and video generation technology are used to generate on-site measured tunnel structures, forming relatively complete tunnel sections and different track structures. Structural layering identification and automatic generation are performed, and long-section track structures are reasonably positioned according to mileage and numbering. For track structures at different locations, the simulation calculation results and the structural 3D model are bound, which can realize the refined management of the track structure and the query of its performance, and analyze the impact of the deformation of the track mechanism on the track structure at different locations.

[0079] Step S9. Use simulation result data combined with deep learning methods to evaluate and predict the service status of the track structure.

[0080] The input data of the track structure service status analysis model are the vehicle dynamic response data and track structure service status data calculated by the vehicle-ballastless track-foundation large deformation coupling analysis model and the digital twin model. Combined with the graded warning threshold, the output data is the judgment data on the safe service status of the track structure. The output data of this step can effectively guide the actual use on site according to the graded warning limit.

[0081] The difference between this part of the track structure service status analysis model and the track structure service status analysis model in step 3 lies in the model data. The data in this part of the model mainly includes the track structure service status data and vehicle dynamics performance data after being processed by deep learning methods. It is a perfection of the process in step 3 and a further display after the data is supplemented.

[0082] (1) The simulation results are combined with the characteristics and change trends of the field measured data, and the measured multi-source heterogeneous data are combined with the deep learning method to predict the data development trend. The development trends of some important parameters are reasonably corrected in combination with the simulation model. According to the predicted data change trend and the vehicle dynamics simulation results, a basic large deformation-vehicle dynamics response mapping mechanism is formed.

[0083] (2) A dynamic vehicle operation discrimination mechanism is established based on input conditions such as large foundation deformation and vehicle speed. Combined with the impact of data changes on the service status of the track structure and the dynamic response of the vehicle, the track structure service status and vehicle dynamic performance are evaluated, and a maintenance plan for the track structure itself and a vehicle speed adjustment plan are proposed for different deformation conditions. Based on the parameters of the tunnel lining deformation conditions and the track structure deformation conditions obtained by the foundation deformation monitoring system, the vehicle operation in the large foundation deformation section is guided. After the foundation deformation exceeds the limit, measures such as vehicle speed control are taken to ensure driving safety and comfort.

[0084] Figure 3It is a technical roadmap for the digital twin method of multivariate heterogeneous data feedback mapping. Under the analysis of limited boundary conditions, existing constitutive relations and contact relations and other calculation assumptions, when arranging points for on-site monitoring, it is necessary to design special measurement points for simulation result verification. According to the on-site monitoring results, the accuracy and deviation of the simulation result calculation are judged, and the simulation results of the model are corrected by a variety of correction calculation methods. The simulation results after feedback are visualized again. At the same time, according to the real-time feedback of the simulation results, it is also necessary to effectively correct and adjust the constitutive relationship of the finite element simulation analysis model. Deep learning and multi-objective optimization theory are used to judge the reasons for the existence of on-site measured data and calculation simulation results. Due to the scale effect between the indoor experimental model and the on-site structure, there may be inaccurate calculation of the constitutive relationship. The main influencing factors of the difference between the simulation results and the measured data, namely the contact relationship and other constitutive relationships set by them, are analyzed. The constitutive relationship with the greatest influence is corrected in real time by the model correction method. Then, the simulation results and the on-site measured data are compared and analyzed again to further optimize and adjust the simulation model.

[0085] Figure 4 It is a method for establishing a simulation result library that combines finite element fast simulation calculation technology and deep learning data analysis methods. That is, first, the finite element simulation fast simulation method is used to calculate the hourly simulation result data, and then it is imported into the deep learning network. Through node displacement prediction, node degree of freedom reduction and a series of matrix fast calculation methods, the deep learning calculation results are verified through on-site monitoring data. If the accuracy verification fails, the model correction technology is used to calibrate the model and re-import it into the deep learning network for calculation. After passing the accuracy verification, the hourly and minute-level simulation result data are imported into the simulation model full-process result library.

[0086] Figure 5 and Figure 6 It is the simulation analysis result of the wheel-rail vertical and lateral forces in the section with large foundation deformation, which shows the change of the wheel-rail vertical and lateral forces with the running time. It can be clearly seen that the wheel-rail vertical force fluctuates significantly when passing through the section with large foundation deformation.

[0087] Figure 7 and Figure 8 It is the simulation analysis result of the vertical displacement and vertical acceleration of the rail at different positions. The monitoring of the rail displacement and acceleration can be used to monitor and analyze the changes in the track structure at the large deformation of the foundation.

[0088] In summary, the embodiment of the present invention proposes a digital twin method of multi-source heterogeneous data feedback mapping for long-term service status assessment of track structures, which can realize real-time acquisition of the service status of the track structure and prediction of its long-term development performance, so as to accurately reflect and visualize in real time the stress characteristics of the track structure under the influence of large deformation of the foundation, and at the same time, make real-time prediction and reasonable evaluation of changes in the stress state of the track structure.

[0089] The present invention proposes a digital twin method of multi-source heterogeneous data feedback mapping for long-term service status evaluation of track structures. It can be combined with a three-dimensional nonlinear displacement application method, and finite element software and multi-body dynamics software are used to simulate the damage of ballastless track structures under different large foundation deformation conditions and analyze the vehicle dynamic response, which provides an effective basis for the service performance evaluation and vehicle dynamic response evaluation of ballastless tracks in large foundation deformation sections.

[0090] Those skilled in the art can understand that the accompanying drawings are only schematic diagrams of an embodiment, and the modules or processes in the accompanying drawings are not necessarily required to implement the present invention.

[0091] It can be known from the description of the above implementation methods that those skilled in the art can clearly understand that the present invention can be implemented by means of software plus a necessary general hardware platform. Based on such an understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which can be stored in a storage medium such as ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in the various embodiments of the present invention or certain parts of the embodiments.

[0092] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device or system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment. The device and system embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0093] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed by the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A digital twin method for multi-source heterogeneous data feedback mapping for long-term service status assessment of rail structures, characterized in that: include: The ballastless track structure is monitored on site through sensors to obtain the structural deformation, displacement, stress, strain, crack and gap width data of the ballastless track structure, wherein the structural deformation and displacement data are used as input data of the finite element model, and the stress, strain, crack and gap width data are used as model correction parameters and as monitoring data to establish a monitoring data model; A vehicle-ballastless track-foundation large deformation coupling analysis model is constructed by using ABAQUS finite element software and multi-body dynamics software, and the basic deformation parameter data of the ballastless track structure and the parameter data of the track structure and the vehicle itself are input into the vehicle-ballastless track-foundation large deformation coupling analysis model, and the vehicle-ballastless track-foundation large deformation coupling analysis model outputs vehicle dynamic response data; Establishing a finite element simulation model for track structure service status analysis, inputting structural deformation, displacement data and foundation deformation parameter data of the ballastless track structure into the track structure service status simulation analysis model, outputting track structure service status data under different foundation deformations, and determining the graded warning threshold for safe service of the track structure according to the track structure service status evaluation index; The digital twin model includes a finite element model, a virtual model and an actual model. The finite element model in the digital twin model is calculated and corrected by a simulation correction method based on coupled feedback of simulation calculation results and analysis data, so as to achieve mutual correspondence between the virtual model and the actual model in the digital twin model. The vehicle-track coupling dynamics model and the track structure service status assessment model are used as simulation models, and the vehicle dynamics response data output by the vehicle-ballastless track-foundation large deformation coupling analysis model and the track structure service status data under different foundation deformations output by the track structure service status analysis model are used as simulation result data; Inputting the measured transverse and vertical displacement monitoring data of the rails, the measured surface stress data of the track structure and the displacement data of the track structure, the measured data of the gap and crack width of the track structure, and the measured data of vehicle dynamics and track structure service status into the digital twin model, combining the vehicle dynamics simulation data and the track structure service status simulation data under different foundation deformations, and using the digital twin model to output the corrected track structure service status data and vehicle dynamics response data; Use deep learning to predict and expand the revised digital twin model data, and present model calculation results in minutes; The deep learning method is used to predict and linearly interpolate the simulation calculation results of the model. The simulation calculation results are predicted and linearly interpolated using the LSTM neural network to achieve full process coverage and full space analysis of the foundation deformation results from the time domain perspective. Use digital twin model data combined with deep learning methods to evaluate and predict the service status of track structures; The input data of the track structure service status analysis and evaluation model are the modified vehicle dynamics simulation data and track structure service status data calculated by the digital twin model. Combined with the graded warning threshold calculated by simulation in step 3, the output data is the judgment data on the safe service status of the track structure, that is, the safe service status of the track structure is reasonably divided.

2. The digital twin method for long-term service status evaluation of ballastless track structure according to claim 1 is characterized in that: The method further comprises: Realize the miniaturization of digital twin model data, establish a three-dimensional digital model of the basic large deformation section, establish a data display platform, and realize the real-time display of digital twin data.

3. The digital twin method for long-term service status evaluation of ballastless track structure according to claim 1 is characterized in that: The vehicle-ballastless track-foundation large deformation coupling analysis model is constructed by using ABAQUS finite element software and multi-body dynamics software, and the basic deformation parameter data of the ballastless track structure and the parameter data of the track structure and the vehicle itself are input into the vehicle-ballastless track-foundation large deformation coupling analysis model. The vehicle-ballastless track-foundation large deformation coupling analysis model outputs vehicle dynamic response data, including: According to the stress, strain and deformation results of the ballastless track structure calculated by ABAQUS software, the finite element simulation model was modified using field measured data, and the data for the next step of deep learning calculation was generated in combination with the digital twin model. In terms of obtaining simulation data for the computational simulation database, the finite element software ABAQUS and the multi-body dynamics software SIMPACK and UM were used, and the calculation efficiency was comprehensively considered to establish a vehicle-ballastless track-foundation large deformation coupling analysis model; In the vehicle-ballastless track-foundation large deformation coupling analysis model, the rails are established as beam units or solid units as track structure facilities that are in direct contact with the moving train; the fasteners are implemented using connector units to simulate the stiffness and damping of the fastener system; the track plate, self-compacting concrete and base plate are all simulated using solid units, while the tunnel lining is simulated using solid units, and some dimensions are simplified. The external rock mass is treated with grounding springs or infinite element method to consider the supporting effect of the geotechnical structure, and the contact relationship between the various layers of the structure and the stress conditions of the layered structure can be considered; The boundary conditions and load loading boundaries of the large deformation coupling analysis model of the vehicle-ballastless track-foundation are determined. The structural deformation data and stress deformation data of the ballastless track structure collected by the sensor are used as the calculation initial conditions in the calculation simulation database. The form and basic parameters of the foundation deformation are comprehensively judged based on the data of each part, and the wavelength and wave depth of the deformation are fitted for the large deformation condition of the foundation. The change form of the fault, the basic deformation parameters of the fault and the fault angle of the line itself are analyzed for the fault offset, and then the forced displacement is applied by a three-dimensional nonlinear curve to simulate the large deformation of the foundation. According to the basic parameters of the vehicle in the large deformation section, the vehicle's motion model is established, and the wheel-rail contact relationship is established according to the Hertz nonlinear contact theory. The calculation of the wheel-rail vertical and lateral forces and the vehicle body acceleration evaluation indicators is realized. The vehicle dynamics response is compared and verified using existing research reports, and then the vehicle dynamics performance response under different basic large deformation amplitudes, different action positions and different vehicle speeds is analyzed and judged.

4. The digital twin method for long-term service status evaluation of ballastless track structure according to claim 1, characterized in that: The establishment of the track structure service status analysis model comprises: According to the elastic-plastic damage parameters of the material, combined with indoor experimental data, an elastic-plastic damage model is established. The elastic-plastic damage model is used to analyze the evolution law of the performance of the structural materials of the ballastless track structure under long-term operation conditions, and the changes in the mechanical properties of the ballastless track structure itself under long-term operation conditions are analyzed. The structural damage parameters of the ballastless track structure under long-term operation conditions are assigned, and damage evaluation criteria are used for different positions of the ballastless track structure. The sub-model technology and the extended finite element method, the cohesive force model and the fracture mechanics analysis method are used to calculate the surface cracks of the track plate and the location of the structural separation. The simulation calculation of the ballastless track structure subjected to large foundation deformation within the large foundation deformation range is carried out to form a damage result library of the ballastless track structure under different large foundation deformations. The damage conditions of the ballastless track structure under the influence of different large foundation deformation amplitudes are fitted using a deep learning method, and the results of different damage conditions of the ballastless track structure are fully covered and automatically generated.

5. The digital twin method for long-term service status assessment of ballastless track structure according to claim 1, characterized in that: The method of using simulation correction based on coupled feedback of simulation calculation results and analysis data to calculate and correct the finite element model in the digital twin model, and realize the mutual correspondence between the virtual model and the actual model in the digital twin model, includes: Under the analysis of limited boundary conditions, existing constitutive relations and contact relationship calculation assumptions, the track structure force is monitored based on the field measured data, and the foundation large deformation curve and amplitude at this time are calculated based on the mapping relationship. The foundation deformation curve and amplitude in the monitoring data of the foundation deformation monitoring sensor at this time are compared, and the simulation model itself is reasonably modified in combination with the data analysis results of other sensors; According to the real-time feedback of the simulation results, the constitutive relationship of the finite element simulation analysis model is corrected and adjusted. The deep learning and multi-objective optimization theory are used to judge the reasons for the differences between the field measured data and the calculated simulation results. The influencing factors of the differences between the simulation results and the measured data are analyzed, and the model correction method is used to make real-time corrections to the constitutive relationship with the greatest influence, realizing the integrated use of simulation results and field measured results.

6. The digital twin method for long-term service status evaluation of ballastless track structure according to claim 1, characterized in that: The deep learning method described above considers predicting and linearly interpolating the simulation calculation results of the model, and the simulation calculation results are predicted and linearly interpolated using the LSTM neural network, so as to achieve full process coverage and full space analysis of the basic deformation results from the time domain perspective, including: A simulation result library establishment method that combines finite element fast simulation calculation technology and deep learning data analysis methods is developed. Fast calculation and simulation analysis are carried out on the established finite element model. A deep learning method is used to consider predicting and linearly interpolating the simulation calculation results of the model. The simulation results are predicted and linearly interpolated using an LSTM neural network. From the time domain perspective, full-process coverage and full-stage simulation analysis of the foundation deformation results are achieved, thus realizing full-process structural service performance simulation.

Images