Bogie digital twin model construction method and system, electronic device and medium

Abstract

The application relates to the technical field of digital twinning, and discloses a bogie digital twinning model construction method, a bogie digital twinning model construction system, electronic equipment and a medium. The method comprises the following steps: acquiring design parameters of a bogie and bogie operation data collected by sensors under multiple working conditions; creating a three-dimensional geometric model of the bogie, and performing dynamic simulation and structure strength finite element simulation on key components of the bogie under multiple working conditions; training a dynamic related lightweight model and a structure strength related lightweight model by using dynamic simulation data and structure strength simulation data; when operation data is acquired, obtaining bogie operation state data based on the dynamic related lightweight model and the structure strength related lightweight model, and displaying the bogie operation state data through the three-dimensional geometric model. The bogie digital twinning model is quickly driven by lightening the digital model, and the bogie physical entity can be more timely monitored and tracked.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of digital twinning, in particular to a bogie digital twinning model construction method and system, electronic equipment and medium. BACKGROUND

[0002] The statements in this section merely provide background information related to the present application and do not necessarily constitute prior art.

[0003] Digital twinning is a means to realize a virtual space reflecting a real digital mirror of a physical object. Through the establishment of a two-way mapping between the physical space and the virtual space, the behavior of the physical entity can be synchronously perceived, monitored, analyzed and predicted from the virtual space, and finally fed back to the physical entity object to realize its maintenance, modification and optimization. The application of digital twinning has the characteristics of comprehensiveness, systematicness, intuitiveness, real-time and high efficiency.

[0004] At present, the application of digital twinning technology related to rail transit is mainly concentrated in the construction management of lean assembly line in the production workshop, the information visualization of subway station equipment, and the driving of train operation optimization, etc. However, there are few digital twinning models applied to trains or key components. The inventor found that the existing related technology applied to trains and key components mainly focuses on the implementation of digital twinning, but the performance of digital twinning model, such as the prediction accuracy of numerical model and the real-time performance of digital twinning model, is poor, which reduces the prediction accuracy and real-time performance of the bogie running state, and is not conducive to train operation and dispatching. SUMMARY

[0005] In order to overcome the shortcomings of the prior art, the present application provides a bogie digital twinning model construction method, system, electronic equipment and medium. On the basis of constructing a three-dimensional geometric model and a digital model, the digital model is also lightened, realizing the rapid driving of the digital twinning model, which is conducive to more timely monitoring and tracking of the physical entity of the bogie.

[0006] To achieve the above object, one or more embodiments of the present application provide the following technical solutions:

[0007] A bogie digital twinning model construction method, comprising the following steps:

[0008] Obtaining design parameters of the bogie and bogie running data collected by sensors under multiple working conditions;

[0009] Based on the design parameters, a three-dimensional geometric model of the bogie is created, and for multiple working conditions, dynamic simulation and structural strength finite element simulation of key components of the bogie are performed to obtain dynamic simulation data and structural strength simulation data;

[0010] The dynamics-related lightweight model and the structure strength-related lightweight model are trained by using the dynamics simulation data and the structure strength simulation data.

[0011] When the running data is acquired, bogie running state data is obtained based on the dynamics-related lightweight model and the structure strength-related lightweight model, and is displayed through the three-dimensional geometric model.

[0012] Further, the dynamics simulation and the structure strength finite element simulation on the key components of the bogie include:

[0013] According to the three-dimensional geometric model and the running data, dynamics simulation on the key components of the bogie under different working conditions is performed to obtain dynamics simulation data of the bogie.

[0014] According to the three-dimensional geometric model and the dynamics simulation data, structure strength finite element simulation on the key components of the bogie under different working conditions is performed to obtain structure strength simulation data.

[0015] Further, after the dynamics simulation data and the structure strength simulation data are obtained, the dynamics simulation data and the structure strength simulation data are screened respectively to obtain dynamics training data and structure strength training data corresponding to different key components.

[0016] Further, for the dynamics simulation data, data cleaning is performed based on preset evaluation indexes (such as safety, stability, comfort, etc.) to obtain dynamics training data.

[0017] Further, the dynamics-related lightweight model and the structure strength-related lightweight model are trained based on a radial basis function neural network.

[0018] One or more embodiments provide a bogie digital twin model construction system, comprising:

[0019] A data acquisition module is configured to acquire design parameters of a bogie and running data of the bogie collected by sensors under multiple working conditions.

[0020] A model simulation module is configured to create a three-dimensional geometric model of the bogie based on the design parameters, and perform dynamics simulation and structure strength finite element simulation on key components of the bogie under multiple working conditions to obtain dynamics simulation data and structure strength simulation data.

[0021] A lightweight model construction module is configured to train a dynamics-related lightweight model and a structure strength-related lightweight model by using the dynamics simulation data and the structure strength simulation data.

[0022] The digital twin model driving module is configured to obtain bogie running state data based on the dynamics-related lightweight model and the structure strength-related lightweight model when the running data is acquired, and display the bogie running state data through the three-dimensional geometric model.

[0023] Further, the dynamics simulation and the structure strength finite element simulation of the key components of the bogie include:

[0024] According to the three-dimensional geometric model and the running data, the dynamics simulation of the key components of the bogie under different working conditions is performed to obtain dynamics simulation data of the bogie;

[0025] According to the three-dimensional geometric model and the dynamics simulation data, the structure strength finite element simulation of the key components of the bogie under different working conditions is performed to obtain structure strength simulation data.

[0026] Further, after the dynamics simulation data and the structure strength simulation data are obtained, the dynamics simulation data and the structure strength simulation data are screened respectively to obtain dynamics training data and structure strength training data corresponding to different key components.

[0027] Further, for the dynamics simulation data, data cleaning is performed based on preset evaluation indexes (such as safety, stability, comfort, and the like) to obtain dynamics training data.

[0028] Further, the dynamics-related lightweight model and the structure strength-related lightweight model are trained based on a radial basis function neural network.

[0029] One or more embodiments provide an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the processor implements the bogie digital twin model construction method when executing the program.

[0030] One or more embodiments provide a computer readable storage medium having a computer program stored thereon, wherein the program is executable by a processor to implement the bogie digital twin model construction method.

[0031] The above one or more technical solutions have the following beneficial effects:

[0032] By integrating real-time sensing data or historical running data of the key components of the bogie under multiple working conditions, the simulation of the dynamics of the bogie and the structure strength of the key components is realized. Based on the data obtained through the simulation, the dynamics-related lightweight digital model and the structure strength-related lightweight digital model corresponding to each module are trained based on a neural network, thereby saving the simulation time. In addition, when the digital twin model is driven based on real-time running data, the running state of the bogie can be quickly predicted, the real-time performance of the digital twin model is improved, and reference basis is provided for operation and maintenance scheduling. Attached Figure Description

[0033] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0034] Figure 1 This is an overall flowchart of the method for constructing a digital twin model of a bogie in one or more embodiments of the present invention;

[0035] Figure 2 This is a schematic diagram illustrating the principle framework for constructing a digital twin model of a bogie in one or more embodiments of the present invention.

[0036] Figure 3 This is a flowchart illustrating the specific process of constructing a digital twin model of a bogie in one or more embodiments of the present invention.

[0037] Figure 4 This is a flowchart of the SIMPACK simulation analysis of the bogie dynamic performance in one or more embodiments of the present invention;

[0038] Figure 5 This is a flowchart of finite element simulation analysis in one or more embodiments of the present invention;

[0039] Figure 6 This is a flowchart illustrating the training of a numerical model based on a radial basis function neural network in one or more embodiments of the present invention;

[0040] Figure 7 This is a diagram of the radial basis function neural network structure in one or more embodiments of the present invention;

[0041] Figure 8 This is a comparison chart of the actual values ​​of wheel-rail vertical force and the predicted values ​​of the lightweight model in one or more embodiments of the present invention;

[0042] Figure 9 This is a comparison chart of the actual values ​​of wheel-rail lateral forces and the predicted values ​​of the lightweight model in one or more embodiments of the present invention.

[0043] Figure 10 This is a comparison chart of the actual values ​​of the derailment coefficient and the predicted values ​​of the lightweight model in one or more embodiments of the present invention. Detailed Implementation

[0044] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0045] It is to be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of example embodiments according to the present application. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "comprises" and / or "comprising," when used in this specification, specify the presence of stated features, steps, operations, elements, components, and / or groups thereof, but do not preclude the presence or addition of one or more other features, steps, operations, elements, components, and / or groups thereof.

[0046] The embodiments in the present application and the features in the embodiments can be combined with each other in the case of no conflict.

[0047] Embodiment one

[0048] The vehicle structure digital twin is a digital thread driven, multi-disciplinary, multi-physical field, multi-scale, multi-fidelity, multi-probability simulation system, which can use multi-source data such as online sensor monitoring, offline inspection, operation history, to reflect and predict the behavior and performance of the corresponding structure entity in the whole life cycle. Based on this, referring to Figure 1 and Figure 2 The embodiment discloses a bogie digital twin model construction method, comprising the following steps:

[0049] Step 1: Obtain the design parameters of the bogie and the bogie operation data collected by the sensors under multiple working conditions.

[0050] The purpose of obtaining the bogie design parameters is to construct the physical entity of the bogie, which includes the frame, wheel set, axle box bearing, suspension system, braking system, traction system and other key components; the bogie design parameters include structure geometric parameters, process parameters, suspension parameters, physical parameters of each component, material parameters, etc., which can be read according to the design drawings.

[0051] The purpose of obtaining the bogie operation data collected by the sensors is to drive the model. The sensors include various sensors for monitoring the operation state of the bogie, such as acceleration sensors, vibration sensors, temperature sensors, speed sensors, stress sensors, displacement sensors, etc., and the operation data can include historical operation data and real-time operation data.

[0052] Output the key components and system evaluation index related curve chart display.

[0053] Step 2: Based on the design parameters, create a three-dimensional geometric model of the bogie.

[0054] Based on the design parameters, create a three-dimensional geometric model of the bogie. Specifically, based on the obtained design parameters, use three-dimensional modeling software such as SolidWorks, CATIA, etc. to construct the three-dimensional geometric model of each bogie key component, and at the same time establish the assembly model of the bogie.

[0055] Step 3: According to the three-dimensional geometric model and the operation data, dynamic simulation under different working conditions is performed to obtain bogie dynamic simulation data.

[0056] A bogie dynamic model is established by using simpack software, as shown in Figure 3 , specifically including: obtaining main technical parameters of the bogie; simplifying the vehicle system and defining the topological relationship of multi-body elements; establishing a simpack simulation model of the vehicle system, performing simulation based on the set simulation running conditions, and obtaining bogie dynamic simulation data through analysis and evaluation.

[0057] The bogie dynamic simulation data includes lateral force, vertical force and acceleration at key positions of the bogie.

[0058] In this embodiment, the dynamic simulation is a simulation of the entire bogie running state and running safety, and the vehicle operation; the structure strength simulation is a consideration of the structure strength of the key components of the bogie.

[0059] Step 4: According to the three-dimensional geometric model and the dynamic simulation data, structure strength finite element simulation of the key components of the bogie under different working conditions is performed to obtain structure strength simulation data.

[0060] Based on the geometric parameters and the dynamic simulation output of multi-working-condition lateral force, vertical force and acceleration, structure strength finite element simulation of the key components of the bogie is performed, as shown in Figure 5 .

[0061] Specifically, as shown in Figure 4 , first, a bogie finite element simulation model is established based on the bogie three-dimensional geometric model. Specifically, by using three-dimensional models of bogie frames, axle box bearings, wheelsets and suspension systems, material properties are defined and a network is divided, joint simulation is performed by using hypermesh and ansys software, and a finite element simulation model of the bogie is established. Then, based on the operation conditions of each component of the bogie, load is applied, resource calling is performed by using isight software, data flow processing is performed, structure strength simulation of the finite element model of each key component of the bogie is performed according to the output load data, structure dynamic and static strength evaluation is performed, key stress and strain values at key component feature positions of the bogie frame, wheelset, suspension system and axle box bearing are output as feature data, and structure-temperature field coupling evaluation of the axle box bearing is realized, and stress, strain and temperature values of key positions of the bearing are output as feature data.

[0062] Step 5: Dynamic training data and structure strength training data corresponding to different key components are obtained by screening based on the dynamic simulation data and the structure strength simulation data respectively.

[0063] Specifically, first, the number and position of key nodes of the part, the data type, and the sample size of the training model are determined; then, for the dynamic simulation data, data cleaning is performed based on safety, stability, comfort, and other evaluation indexes to obtain dynamic training data, and for the structural strength simulation data, data cleaning is performed to obtain structural strength training data.

[0064] It should be noted here that other evaluation indexes can be selected by those skilled in the art for data cleaning according to actual conditions.

[0065] In order to obtain a lightweight digital twin model subsequently, the embodiment filters based on the dynamic simulation data and the structural strength simulation data as training data samples for constructing a lightweight numerical model. For each numerical model to be trained, the corresponding dynamic simulation data and structural strength simulation data after filtering are respectively split in a ratio of 8:2, wherein 80% of the data is used as a lightweight numerical model training machine learning sample, and 20% is used as a sample data for verifying the digital lightweight model.

[0066] As an example, the main suspension parameters of the vehicle are selected as inputs, and the above suspension parameters are designed for DOE (Design of Experiment) test with 50% of the original parameters as the boundary, so as to select test factors, determine the number of levels of each factor, establish an orthogonal table, calculate each group of indexes, and continue to analyze to reduce simulation conditions while obtaining more simulation information. The suspension parameter input data sample set is generated, and then the output response of each working condition based on the suspension parameter input data sample set is obtained through the SIMPACK dynamics software, i.e., the dynamic simulation data such as wheel axle lateral force, wheel rail vertical force, speed, wheel load reduction rate, and derailment coefficient corresponding to each suspension parameter working condition are calculated.

[0067] The output response data set is divided into training data and test data in a certain ratio, wherein the training data sample is used for training the lightweight model, and the test data sample is used for judging the effect of the lightweight model. Specifically, 80% of the data is used as training data, and 20% is used as test data.

[0068] Step 6: Based on the dynamic training data and the structural strength training data, respectively, the dynamic related lightweight model and the structural strength related lightweight model corresponding to different key components are trained based on a neural network.

[0069] The dynamic related lightweight model includes lightweight models of safety, comfort, and stability corresponding to different key components; and the structural strength related lightweight model includes lightweight models of stress, strain, fatigue, and life corresponding to different key components.

[0070] The lightweight model obtained by training the simulation data is: a bogie structure strength lightweight model, a bogie lateral acceleration lightweight model, a derailment coefficient lightweight model, etc.

[0071] In the embodiment, the neural network adopts an RBF (Radial Basis Function) neural network which has a good approximation effect on a nonlinear function.

[0072] Figure 7 The RBF neural network adopts a 3-layer neural network structure with a single hidden layer, including an input layer (8 neurons), a hidden layer (16 hidden factors), and an output layer (12 neurons). The loss function is The learning rate coefficient is set to 0.05. The RBF neural network center, variance (width), and the weight parameter ω from the hidden layer to the output layer are randomly selected. i The iterative calculation adopts the training method of the RBF neural network weight parameter, and the gradient descent method is adopted in the embodiment. The number of training is determined according to the model effect after training, and the lightweight model meets the error requirement through multiple training.

[0073] After the lightweight model is trained, the model prediction result is verified. Specifically, as shown in Figure 6 , the verification sample in the sample or the processed sensor real-time data and historical operation data are used to verify the lightweight model of each module, the output result is analyzed, and the neural network structure parameter is adjusted to train each module to the required accuracy. With the radial basis function neural network and the iterative optimization of the model, the prediction accuracy and the calculation efficiency are considered.

[0074] Taking the bogie dynamics lightweight model neural network construction as an example, the x1-x8 axle box vibration acceleration is taken as the input, and the y1-y12 axle box vibration acceleration corresponding to the wheel rail force and the derailment coefficient of the wheel are taken as the output, so that the lightweight model (such as the dynamics wheel rail force lightweight model) can be trained. The dynamics wheel rail force lightweight model is driven based on the verification sample data not used for model training, and part of the output result curves are as shown in Figure 8 、 Figure 9 and Figure 10 . By comparing the actual value curve and the simulation value curve, it can be seen that the prediction value curve is basically consistent with the actual value curve, and the error is small, so it can be determined that the dynamics wheel rail force lightweight model module is completed.

[0075] The error statistics of the plurality of lightweight models based on the entire bogie model and the actually measured data are shown in Table 1. The maximum error of the lightweight model output of the motor train at a speed of 300 km / h is 6.17%, and the overall error range is within 10%, which can meet the requirements of most projects.

[0076] Table 1 Error statistics table of the actual value and the lightweight model prediction value of the motor train at a speed of 300 km / h

[0077]

[0078] Preferably, for different accuracy requirements of different projects, the lightweight model can be further optimized, such as increasing the training data samples, increasing the sample data accuracy, adjusting the neural network structure parameters, etc., which can further improve the results of the lightweight model.

[0079] Step 7: When obtaining the running data, the bogie running state data is obtained based on the dynamics related lightweight model and the structure strength related lightweight model, and is displayed through the three-dimensional geometric model.

[0080] The running data can be any one of the following: real-time running data obtained by sensors, historical running data, and running data obtained under simulated working conditions. The running data is used as the input of the verified lightweight digital model, the output features are processed one by one with the key nodes of the three-dimensional geometric model, and the corresponding evaluation indexes are output, realizing the construction of the digital twin model of the bogie. The evaluation indexes can be displayed through curves or charts. Based on the existing simulation data and historical running data, the virtual driving of the bogie digital twin model is helpful to verify the running state of the bogie.

[0081] The embodiment also performs lightweight processing on the three-dimensional geometric model of the bogie. The numerical analysis method provides calculation accuracy at the cost of the spatial discrete scale of the component model. The digital prototype construction method mainly based on numerical analysis method and model reduction technology well solves the contradiction between calculation accuracy and calculation speed. The digital prototype method is used to process the key component model and assembly model of the bogie. Specifically, the digital prototype technology is used to completely describe and express the key component model and assembly model in the three-dimensional geometric model of the bogie. The key parameters output by the digital model are reflected through feature data, and the stress, temperature, displacement and other states of the components are displayed or alarmed through rendering technology. Thus, the real-time driving of the bogie digital twin model is realized, and the real-time data of the bogie can be presented through intuitive graphics, improving the efficiency of information sharing and data analysis, and providing a reference for operation and dispatching.

[0082] The three-dimensional geometric model corresponds to a physical bogie entity, the dynamics-related lightweight model and the structure strength-related lightweight model correspond to a digital bogie model, and the physical bogie entity and the digital bogie model jointly constitute a digital twin model.

[0083] In order to realize real-time operation data-driven digital twin model, after the lightweight digital model is trained, first, different lightweight digital models are packaged by modules, and through the combination and packaging of multiple lightweight digital twin models, various project digital twin models with different requirements can be quickly assembled.

[0084] Then, the input and output data interfaces of each digital model are set, the data interface types of real-time sensor data and lightweight models are determined, and it is ensured that the data can be normally input and drive the model. Specifically, the business logic processing is programmed by using Python language and the like, the reduced-order digital model that meets the requirements and is trained is set in the interface type by calling the related function, is packaged in the "ROM" package, and the data input and output interfaces are determined by program writing to realize the connection between the characteristic data key nodes of the display model and the characteristic data key nodes of the digital model, so that the output of the digital model can correspond to the evaluation index, image display function and the like of the display model, and the packaging of the digital twin model is completed.

[0085] When the bogie key part digital twin model is arranged on the corresponding display terminal, the functions that can be realized include but are not limited to vehicle operation safety evaluation, which can output derailment coefficient, wheel load reduction rate, wheel-rail lateral force and the like; bogie frame strength evaluation, wheel set structure strength evaluation, suspension system structure strength evaluation, axle box bearing structure strength evaluation and bogie state and the like.

[0086] The bogie digital twin model can be used for multi-platform deployment and intuitive display of multi-dimensional driving state.

[0087] Embodiment Two

[0088] The purpose of this embodiment is to provide a bogie digital twin model construction system. The system comprises:

[0089] A data acquisition module is configured to acquire design parameters of a bogie and bogie operation data collected by sensors under multiple working conditions;

[0090] A model simulation module is configured to create a three-dimensional geometric model of the bogie based on the design parameters, and perform dynamics simulation and structure strength finite element simulation on key components of the bogie under multiple working conditions to obtain dynamics simulation data and structure strength simulation data;

[0091] The lightweight model construction module is configured to train a dynamics-related lightweight model and a structure strength-related lightweight model based on the dynamics simulation data and the structure strength simulation data.

[0092] The digital twin model driving module is configured to obtain bogie running state data based on the dynamics-related lightweight model and the structure strength-related lightweight model when the running data is obtained, and display the bogie running state data through the three-dimensional geometric model.

[0093] Embodiment Three

[0094] An object of the embodiment is to provide an electronic device.

[0095] An electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the bogie digital twin model construction method as described in Embodiment One when executing the program.

[0096] Embodiment Four

[0097] An object of the embodiment is to provide a computer-readable storage medium.

[0098] A computer-readable storage medium has a computer program stored thereon, and the program is executable on a processor to implement the bogie digital twin model construction method as described in Embodiment One.

[0099] The steps and methods involved in Embodiments Two to Four correspond to Embodiment One, and the specific embodiments can be referred to the relevant description in Embodiment One.

[0100] One or more of the above embodiments integrates the finite element simulation technology of dynamics and structure strength, and combines the RBF neural network algorithm, to realize the construction of the digital twin model of the dynamics and the structure strength of the key components of the bogie based on the historical running data, thereby realizing the functions of the state monitoring and tracking of the digital twin model of the key components of the bogie driven by the real-time sensing data or the historical running data of the bogie. Meanwhile, the rapid simulation of the running conditions or the structure strength based on the digital twin model of the key components of the bogie can be realized, accurate simulation results can be obtained, and the simulation calculation time can be saved. In addition, the digital twin model construction of other components of the rail transit vehicle is provided as a reference.

[0101] Through integrating a large amount of simulation data and historical operation data made by the existing rail transit industry, the training and verification of the digital twin model can be virtually driven, and the selection of the simulation tool does not have specific requirements, and engineers do not have to learn new simulation software again, only need to process the simulation result data and then train the model, which is relatively easy to realize and low in cost. Based on the existing real-time sensor data, dynamic or structural strength simulation data and historical operation data driving the bogie digital twin model, the multiplicity and value of the simulation data in the rail transit industry can be improved.

[0102] Those skilled in the art should understand that the above-mentioned modules or steps of the present application can be realized by a general computer device, alternatively, they can be realized by program codes executable by a computing device, so that they can be stored in a storage device and executed by a computing device, or they can be respectively manufactured into individual integrated circuit modules, or a plurality of modules or steps among them can be manufactured into a single integrated circuit module. The present application is not limited to any specific combination of hardware and software.

[0103] Although the specific embodiments of the present application are described above in combination with the drawings, the present application is not limited to the scope of the drawings, and those skilled in the art should understand that various modifications or changes made on the basis of the technical solutions of the present application without creative labor are still within the protection scope of the present application.

Claims

1. A method for constructing a digital twin model of a bogie, characterized in that, Includes the following steps: Obtain the bogie's design parameters, as well as the bogie's operating data collected by sensors under various working conditions; Based on the design parameters, a three-dimensional geometric model of the bogie is created, and dynamic simulation and structural strength finite element simulation are performed on key components of the bogie under various working conditions to obtain dynamic simulation data and structural strength simulation data. Using the aforementioned dynamic simulation data and structural strength simulation data, a dynamic-related lightweight model and a structural strength-related lightweight model were trained. When acquiring operational data, the bogie operational status data is obtained based on the dynamic-related lightweight model and the structural strength-related lightweight model, and displayed through the three-dimensional geometric model; The three-dimensional geometric model corresponds to the bogie physical entity, and the dynamic-related lightweight model and the structural strength-related lightweight model correspond to the bogie digital model. The bogie physical entity and the digital model together form a digital twin model. After the lightweight digital model is trained, different lightweight digital models are packaged into modules. By combining and packaging multiple lightweight digital twin models, real-time running data-driven digital twin models and rapid assembly of digital twin models can be achieved.

2. The method for constructing a digital twin model of a bogie as described in claim 1, characterized in that, Dynamic simulation and structural strength finite element simulation of key bogie components include: Based on the three-dimensional geometric model and operating data, dynamic simulations of key bogie components under different working conditions are performed to obtain bogie dynamic simulation data. Based on the three-dimensional geometric model and dynamic simulation data, finite element simulations of the structural strength of key bogie components under different working conditions were performed to obtain structural strength simulation data.

3. The method for constructing a digital twin model of a bogie as described in claim 1 or 2, characterized in that, After obtaining the dynamic simulation data and structural strength simulation data, the dynamic simulation data and structural strength simulation data are further filtered to obtain dynamic training data and structural strength training data for different key components.

4. The method for constructing a digital twin model of a bogie as described in claim 3, characterized in that, For dynamic simulation data, data cleaning is performed based on preset evaluation indicators to obtain dynamic training data.

5. The method for constructing a digital twin model of a bogie as described in claim 1, characterized in that, The dynamic-related lightweight model and the structural strength-related lightweight model are trained based on radial basis function neural networks.

6. A bogie digital twin model construction system, characterized in that, include: The data acquisition module is used to acquire the bogie's design parameters and bogie operating data collected by sensors under various working conditions. The model simulation module is used to create a three-dimensional geometric model of the bogie based on the design parameters, and to perform dynamic simulation and structural strength finite element simulation of key components of the bogie under various working conditions, so as to obtain dynamic simulation data and structural strength simulation data. The lightweight model construction module is used to train a dynamic-related lightweight model and a structural strength-related lightweight model using the dynamic simulation data and structural strength simulation data. The digital twin model driving module is used to obtain bogie operating status data based on the dynamic-related lightweight model and the structural strength-related lightweight model when acquiring operating data, and to display it through the three-dimensional geometric model; The three-dimensional geometric model corresponds to the bogie physical entity, and the dynamic-related lightweight model and the structural strength-related lightweight model correspond to the bogie digital model. The bogie physical entity and the digital model together form a digital twin model. After the lightweight digital model is trained, different lightweight digital models are packaged into modules. By combining and packaging multiple lightweight digital twin models, real-time running data-driven digital twin models and rapid assembly of digital twin models can be achieved.

7. The bogie digital twin model construction system as described in claim 6, characterized in that, Dynamic simulation and structural strength finite element simulation of key bogie components include: Based on the three-dimensional geometric model and operating data, dynamic simulations of key bogie components under different working conditions are performed to obtain bogie dynamic simulation data. Based on the three-dimensional geometric model and dynamic simulation data, finite element simulations of the structural strength of key bogie components under different working conditions were performed to obtain structural strength simulation data.

8. The bogie digital twin model construction system as described in claim 6 or 7, characterized in that, After obtaining the dynamic simulation data and structural strength simulation data, the dynamic simulation data and structural strength simulation data are further filtered to obtain dynamic training data and structural strength training data for different key components.

9. The bogie digital twin model construction system as described in claim 8, characterized in that, For dynamic simulation data, data cleaning is performed based on preset evaluation indicators to obtain dynamic training data.

10. The bogie digital twin model construction system as described in claim 6, characterized in that, The dynamic-related lightweight model and the structural strength-related lightweight model are trained based on radial basis function neural networks.

11. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the bogie digital twin model construction method as described in any one of claims 1-5.

12. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the bogie digital twin model construction method as described in any one of claims 1-5.

Images