Battery power system digital twin method and system applied to new energy heavy truck

Abstract

The application discloses a battery power system digital twin method and system applied to a new energy heavy truck, and the method comprises the following steps: acquiring basic data; obtaining a multi-scale mapping model of the battery power system based on the basic data; adding a data-driven model to extract information from real-time monitoring data of the battery power system and supplement the multi-scale mapping model to construct a digital twin model of the battery power system; abstracting sensor data of the battery power system into a network topology graph for space-time relationship multi-scale modeling through a graph neural network-long short-term memory neural network structure; and perceiving running state information of the battery power system by using the constructed digital twin model of the battery power system; and constructing a deep semantic matching fusion model of an incomplete multi-modal multi-physical field by using a cross-modal deep semantic matching mechanism for multi-modal data contained in the perceived running state information.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a digital twin method and system for the battery power system of new energy heavy trucks. Background Technology

[0002] In the field of battery power systems for new energy heavy-duty trucks, existing technologies typically employ rule-based energy management strategies and simplified battery models for system control. For example, traditional methods construct cell-level electrochemical models solely by monitoring apparent parameters such as battery voltage and temperature, and allocate power source output based on fixed thresholds. While this approach achieves basic energy management functions, it suffers from significant technical shortcomings: firstly, because it fails to consider the dynamic impact of multi-physical field coupling (such as electrochemical, temperature, and stress fields) on battery performance, the model cannot accurately reflect the actual state of the battery under complex operating conditions; secondly, the lack of a real-time data-driven model update mechanism makes it difficult to adapt to the nonlinear characteristics of new energy heavy-duty trucks operating under multiple conditions. These shortcomings result in low energy utilization efficiency and lag in system response when dealing with dynamic environmental changes, failing to meet the high energy efficiency and high reliability requirements of new energy heavy-duty trucks. Summary of the Invention

[0003] This invention provides a digital twin method and system for battery power systems applied to new energy heavy trucks, which is capable of...

[0004] One embodiment of the present invention provides a digital twin method for a battery power system applied to new energy heavy trucks, comprising the following steps:

[0005] Acquire basic data for the digital twin of the battery power system applied to new energy heavy trucks. The basic data includes: full-element parameter information of the battery power system of new energy heavy trucks under the multi-physics field coupling state of electrochemical field, temperature field and stress field, as well as multi-scale data from battery cells to modules, packs, controllers and drive motors.

[0006] Based on the aforementioned fundamental data, an electrochemical-thermal coupling model is constructed from the cell to the module scale of the battery power system. A thermo-mechanical coupling model is constructed at the cell scale considering thermal stress and thermal expansion behavior. A geometric model is established and optimized from the module and pack scales of the battery power system. At the same time, a dynamic model of the battery power system is established to achieve a comprehensive abstraction of the battery power system and obtain a multi-scale image model of the battery power system.

[0007] A data-driven model is incorporated to extract information from the real-time monitoring data of the battery power system. The extracted information is then added to the multi-scale imaging model to construct a digital twin model of the battery power system.

[0008] By using a graph neural network-long short-term memory neural network structure, the sensor data of the battery power system is abstracted into a network topology graph for multi-scale modeling of spatiotemporal relationships. The constructed digital twin model of the battery power system is then used to perceive the operating status information of the battery power system.

[0009] By utilizing a cross-modal deep semantic matching mechanism, a shared feature subspace between modalities is constructed for the multimodal data contained in the perceived operational state information, and an invariant graph regularization factor is designed to establish a deep semantic matching fusion model for incomplete multimodal and multi-physics fields, thereby reducing semantic bias.

[0010] As an improvement to the above scheme, based on the aforementioned basic data, an electrochemical-thermal coupling model is constructed from the cell to the module scale of the battery power system. A thermo-mechanical coupling model is constructed at the cell scale, considering thermal stress and thermal expansion behavior. Geometric models are established and optimized at the module and overall pack scales of the battery power system. Simultaneously, a dynamic model of the battery power system is established, achieving a comprehensive abstraction of the battery power system and obtaining a multi-scale imaging model of the battery power system, including:

[0011] Cell-to-module scale modeling: At the cell scale, an electrochemical-thermal model is established based on the P2D model according to the energy conservation equation, resulting in the cell-scale electrochemical-thermal model; at the single-cell scale, a power battery thermo-mechanical coupling model is constructed based on the thermal stress and thermal expansion behavior caused by the temperature gradient during battery operation, resulting in the single-cell-scale thermo-mechanical coupling model.

[0012] Modeling at the module and overall scale: Based on the electrochemical-thermal model at the cell scale and the thermo-mechanical coupling model at the single cell scale, a three-dimensional CAD model of the power battery is constructed using industrial modeling software to accurately express its key structural parameters, constraints and positioning relationships between single cells and modules, and obtain a three-dimensional CAD model; and geometric optimization is performed using PixyzStudio to obtain an optimized geometric model;

[0013] Dynamic modeling: Using the optimized geometric model as input, and combining the mechanical geometry, material parameters, force analysis, and velocity load information of the power battery pack, motor, inverter, and drive control unit, a dynamic model containing mechanical geometry, material parameters, force analysis, and velocity load information is established.

[0014] As an improvement to the above solution, the addition of a data-driven model involves extracting information from the real-time monitoring data of the battery power system and supplementing the extracted information into the multi-scale imaging model to construct a digital twin model of the battery power system, including:

[0015] Model system fusion: The multi-scale image model obtained in the multi-scale modeling step is fused with the data-driven model to form the model system before fusion;

[0016] Real-time data processing: A data-driven approach is used to extract information from real-time monitoring data. The extracted information is then added to the multi-scale imaging model in the pre-fusion model system to form a supplemented multi-scale imaging model. The real-time monitoring data includes sensor data from the physical layer and simulation data from the virtual layer.

[0017] Digital twin construction: Based on the supplemented multi-scale imaging model, a multi-scale digital twin model of the battery power system is constructed to achieve data-driven model fusion.

[0018] As an improvement to the above solution, the method of abstracting the sensor data of the battery power system into a network topology graph using a graph neural network-long short-term memory neural network structure for multi-scale modeling of spatiotemporal relationships, and using the constructed digital twin model of the battery power system to perceive the operating status information of the battery power system, includes:

[0019] Data abstraction processing: The sensor data is abstracted into a network topology diagram of the sensor data time-series accumulation process, and the connections between nodes are defined as sequential events, forming a network topology diagram of the sensor data;

[0020] Dynamic graph construction: Based on the network topology graph of the sensor data, a time-series dynamic graph structure for fusing battery-motor-braking characteristic information is constructed to obtain a time-series dynamic graph containing battery-motor-braking characteristics;

[0021] Feature learning processing: The structural and temporal information of the time-series dynamic graph is integrated using a graph neural network-long short-term memory neural network structure. A hierarchical feature aggregation method is designed to learn feature aggregators in neighborhoods at different depths, so as to realize the flow of information from higher depths to nodes and obtain state perception results.

[0022] As an improvement to the above scheme, the method utilizes a cross-modal deep semantic matching mechanism to construct a shared feature subspace among modalities for the multimodal data contained in the perceived operational state information, and designs an invariant graph regularization factor to establish a deep semantic matching fusion model for incomplete multimodal and multi-physics fields, thereby reducing semantic bias. This includes:

[0023] Cross-modal modeling: A cross-modal deep semantic matching mechanism is adopted, and an incomplete cross-modal deep semantic matching fusion model is established through the multi-level and multi-scale nonlinear correlation between modal data to obtain the initial fusion model;

[0024] Feature space construction: Based on the initial fusion model, a shared feature subspace between modalities is constructed, and the sharing of incomplete cross-modal data is learned to form a shared feature subspace;

[0025] Regularization process: Invariant graph regularization factors are designed in the shared feature subspace to ensure the local similarity characteristics of each modality data, and the regularized shared subspace is obtained;

[0026] Objective function establishment: A new objective function is established based on the regularized shared subspace to form a complete multimodal data fusion model, thereby reducing semantic bias.

[0027] Another embodiment of the present invention provides a digital twin system for the battery power system of new energy heavy trucks, comprising:

[0028] The acquisition module is used to acquire the basic data of the digital twin of the battery power system applied to new energy heavy trucks. The basic data includes: full-element parameter information of the battery power system of new energy heavy trucks under the multi-physics field coupling state of electrochemical field, temperature field and stress field, as well as multi-scale data from battery cells to modules, packs, controllers and drive motors.

[0029] The first construction module is used to construct an electrochemical-thermal coupling model from the cell to the module scale of the battery power system based on the basic data, construct a thermo-mechanical coupling model at the cell scale considering thermal stress and thermal expansion behavior at the cell scale, establish and optimize a geometric model from the module and pack scales of the battery power system, and simultaneously establish a dynamic model of the battery power system to achieve a comprehensive abstraction of the battery power system and obtain a multi-scale image model of the battery power system.

[0030] The second construction module is used to add a data-driven model, extract information from the real-time monitoring data of the battery power system, supplement the extracted information into the multi-scale imaging model, and construct a digital twin model of the battery power system.

[0031] The third construction module is used to abstract the sensor data of the battery power system into a network topology graph through a graph neural network-long short-term memory neural network structure to perform multi-scale modeling of spatiotemporal relationships, and to use the constructed digital twin model of the battery power system to perceive the operating status information of the battery power system.

[0032] The fourth construction module is used to utilize a cross-modal deep semantic matching mechanism to construct a shared feature subspace among modalities for the multimodal data contained in the perceived operational state information, and to design an invariant graph regularization factor to establish a deep semantic matching fusion model for incomplete multimodal and multi-physics fields, thereby reducing semantic bias.

[0033] As an improvement to the above solution, the first construction module is specifically used for:

[0034] Cell-to-module scale modeling: At the cell scale, an electrochemical-thermal model is established based on the P2D model according to the energy conservation equation, resulting in the cell-scale electrochemical-thermal model; at the single-cell scale, a power battery thermo-mechanical coupling model is constructed based on the thermal stress and thermal expansion behavior caused by the temperature gradient during battery operation, resulting in the single-cell-scale thermo-mechanical coupling model.

[0035] Modeling at the module and overall scale: Based on the electrochemical-thermal model at the cell scale and the thermo-mechanical coupling model at the single cell scale, a three-dimensional CAD model of the power battery is constructed using industrial modeling software to accurately express its key structural parameters, constraints and positioning relationships between single cells and modules, and obtain a three-dimensional CAD model; and geometric optimization is performed using PixyzStudio to obtain an optimized geometric model;

[0036] Dynamic modeling: Using the optimized geometric model as input, and combining the mechanical geometry, material parameters, force analysis, and velocity load information of the power battery pack, motor, inverter, and drive control unit, a dynamic model containing mechanical geometry, material parameters, force analysis, and velocity load information is established.

[0037] As an improvement to the above solution, the second building module is specifically used for:

[0038] Model system fusion: The multi-scale image model obtained in the multi-scale modeling step is fused with the data-driven model to form the model system before fusion;

[0039] Real-time data processing: A data-driven approach is used to extract information from real-time monitoring data. The extracted information is then added to the multi-scale imaging model in the pre-fusion model system to form a supplemented multi-scale imaging model. The real-time monitoring data includes sensor data from the physical layer and simulation data from the virtual layer.

[0040] Digital twin construction: Based on the supplemented multi-scale imaging model, a multi-scale digital twin model of the battery power system is constructed to achieve data-driven model fusion.

[0041] As an improvement to the above solution, the third building module is specifically used for:

[0042] Data abstraction processing: The sensor data is abstracted into a network topology diagram of the sensor data time-series accumulation process, and the connections between nodes are defined as sequential events, forming a network topology diagram of the sensor data;

[0043] Dynamic graph construction: Based on the network topology graph of the sensor data, a time-series dynamic graph structure for fusing battery-motor-braking characteristic information is constructed to obtain a time-series dynamic graph containing battery-motor-braking characteristics;

[0044] Feature learning processing: The structural and temporal information of the time-series dynamic graph is integrated using a graph neural network-long short-term memory neural network structure. A hierarchical feature aggregation method is designed to learn feature aggregators in neighborhoods at different depths, so as to realize the flow of information from higher depths to nodes and obtain state perception results.

[0045] As an improvement to the above solution, the fourth building module is specifically used for:

[0046] Cross-modal modeling: A cross-modal deep semantic matching mechanism is adopted, and an incomplete cross-modal deep semantic matching fusion model is established through the multi-level and multi-scale nonlinear correlation between modal data to obtain the initial fusion model;

[0047] Feature space construction: Based on the initial fusion model, a shared feature subspace between modalities is constructed, and the sharing of incomplete cross-modal data is learned to form a shared feature subspace;

[0048] Regularization process: Invariant graph regularization factors are designed in the shared feature subspace to ensure the local similarity characteristics of each modality data, and the regularized shared subspace is obtained;

[0049] Objective function establishment: A new objective function is established based on the regularized shared subspace to form a complete multimodal data fusion model, thereby reducing semantic bias.

[0050] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0051] This invention constructs a multi-physics coupled, multi-scale digital twin model, combines a data-driven approach to achieve dynamic model updates, and utilizes spatiotemporal feature fusion and cross-modal semantic matching techniques to process multi-source heterogeneous data. Specifically, it first collects all-element parameters and multi-scale data of the multi-physics field (electrochemical field, temperature field, stress field) of the battery power system to construct a multi-level physical model encompassing the entire battery pack from the cell level. Then, it supplements the model with real-time monitoring data through a data-driven model, improving model accuracy. Next, it employs a graph neural network-long short-term memory neural network structure to process the spatiotemporal relationships of sensor data, achieving precise perception of operational status. Finally, it establishes a shared feature subspace and introduces a regularization factor through a cross-modal semantic matching mechanism to address the problem of incomplete multimodal data. This technical concept, through a hierarchical model construction and data fusion method, overcomes the limitations of traditional models that rely solely on apparent parameters and lack dynamic adaptability, achieving comprehensive, high-precision simulation and real-time optimization of the battery power system. This invention addresses the technical shortcomings of traditional methods that cannot dynamically reflect the complex operating conditions of batteries by using multi-physics coupling modeling and real-time data-driven mechanisms. It achieves accurate mapping and real-time optimization of the multi-scale operating states of the power system of new energy heavy truck batteries, significantly improving the system's energy utilization efficiency and dynamic response capability. Attached Figure Description

[0052] Figure 1 This is a flowchart illustrating a digital twin method for a battery power system applied to a new energy heavy truck, according to an embodiment of the present invention.

[0053] Figure 2 This is a schematic diagram of the structure of a digital twin system for a battery power system applied to a new energy heavy truck, provided by an embodiment of the present invention. Detailed Implementation

[0054] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0055] See Figure 1 This is a flowchart illustrating a digital twin method for a battery power system applied to new energy heavy-duty trucks, according to an embodiment of the present invention. The digital twin method for a battery power system applied to new energy heavy-duty trucks includes the following steps:

[0056] S10, acquire the basic data of the digital twin of the battery power system applied to new energy heavy trucks. The basic data includes: full-element parameter information of the battery power system of new energy heavy trucks under the multi-physics field coupling state of electrochemical field, temperature field and stress field, as well as multi-scale data from battery cells to modules, packs, controllers and drive motors.

[0057] S11. Based on the basic data, an electrochemical-thermal coupling model is constructed from the cell to the module scale of the battery power system. A thermo-mechanical coupling model is constructed at the cell scale considering thermal stress and thermal expansion behavior. A geometric model is established and optimized from the module and pack scales of the battery power system. At the same time, a dynamic model of the battery power system is established to achieve a comprehensive abstraction of the battery power system and obtain a multi-scale image model of the battery power system.

[0058] S12, Add a data-driven model to extract information from the real-time monitoring data of the battery power system, supplement the extracted information into the multi-scale imaging model, and construct a digital twin model of the battery power system.

[0059] S13, using a graph neural network-long short-term memory neural network structure, the sensor data of the battery power system is abstracted into a network topology graph for multi-scale modeling of spatiotemporal relationships, and the constructed digital twin model of the battery power system is used to perceive the operating status information of the battery power system.

[0060] S14. Using a cross-modal deep semantic matching mechanism, a shared feature subspace between modalities is constructed for the multimodal data contained in the perceived operating state information, and an invariant graph regularization factor is designed to establish a deep semantic matching fusion model for incomplete multimodal and multiphysical fields, thereby reducing semantic bias.

[0061] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0062] This invention constructs a multi-physics coupled, multi-scale digital twin model, combines a data-driven approach to achieve dynamic model updates, and utilizes spatiotemporal feature fusion and cross-modal semantic matching techniques to process multi-source heterogeneous data. Specifically, it first collects all-element parameters and multi-scale data of the multi-physics field (electrochemical field, temperature field, stress field) of the battery power system to construct a multi-level physical model encompassing the entire battery pack from the cell level. Then, it supplements the model with real-time monitoring data through a data-driven model, improving model accuracy. Next, it employs a graph neural network-long short-term memory neural network structure to process the spatiotemporal relationships of sensor data, achieving precise perception of operational status. Finally, it establishes a shared feature subspace and introduces a regularization factor through a cross-modal semantic matching mechanism to address the problem of incomplete multimodal data. This technical concept, through a hierarchical model construction and data fusion method, overcomes the limitations of traditional models that rely solely on apparent parameters and lack dynamic adaptability, achieving comprehensive, high-precision simulation and real-time optimization of the battery power system. This invention addresses the technical shortcomings of traditional methods that cannot dynamically reflect the complex operating conditions of batteries by using multi-physics coupling modeling and real-time data-driven mechanisms. It achieves accurate mapping and real-time optimization of the multi-scale operating states of the power system of new energy heavy truck batteries, significantly improving the system's energy utilization efficiency and dynamic response capability.

[0063] As one embodiment, in step S10, comprehensive multi-physics coupling data and multi-scale structural data of the new energy heavy-duty truck battery power system are collected to provide complete basic information for subsequent digital twin modeling. Traditional technologies rely only on battery apparent parameters (such as voltage and temperature) for modeling, which cannot reflect the dynamic coupling effect of internal multi-physics fields (electrochemical field, temperature field, stress field) and the interaction between multi-scale levels, resulting in insufficient model accuracy. This solution breaks through the single-dimensional limitation of traditional models by simultaneously acquiring all-element parameters (such as electrochemical characteristics, thermal distribution, and mechanical stress) and hierarchical data from the cell to the entire pack (including controllers, drive motors, etc.), laying a data foundation for building a high-precision digital twin model, and ultimately achieving accurate mapping and optimization of the battery power system's entire life cycle state.

[0064] Specifically, for multi-physics data acquisition: a distributed sensor network is used to monitor the electrochemical field (e.g., state of charge, internal resistance), temperature field (e.g., cell / module temperature distribution), and stress field (e.g., mechanical deformation, thermal expansion) parameters of the battery power system in real time, covering three operating conditions: dynamic charging and discharging, constant charging, and static storage. Combined with simulation data (e.g., thermal diffusion trends predicted by the virtual layer), a twin dataset is formed, containing both physical layer monitoring data and virtual layer simulation data. For multi-scale structural data acquisition: microscopic electrochemical parameters (e.g., ion diffusion coefficient) are acquired at the cell level; series and parallel connection relationships and thermal management parameters are recorded at the module level; and overall energy flow and mechanical structure parameters are acquired at the entire battery pack level. Industrial modeling software (e.g., CATIA) is used to construct a 3D CAD model of the power battery, accurately representing key structural parameters (e.g., electrode material thickness, module spacing), and surface subdivision and topology correction are performed using Pixyz Studio to ensure a high degree of consistency between the geometric model and the physical entity. For time-series accumulation of sensor data: Sensor data is abstracted into a network topology graph, where nodes represent sensors (such as voltage sensors and temperature sensors), and edges represent data time-series relationships, forming a dynamically spatiotemporally correlated dataset. Dimensionality reduction is performed on multi-source heterogeneous data using methods such as Non-negative Matrix Factorization (NMF) to extract core feature vectors and reduce data redundancy.

[0065] As one embodiment, based on the aforementioned basic data, an electrochemical-thermal coupling model is constructed from the cell to the module scale of the battery power system; a thermo-mechanical coupling model is constructed at the cell scale considering thermal stress and thermal expansion behavior; a geometric model is established and optimized at the module and overall pack scales of the battery power system; and a dynamic model of the battery power system is simultaneously established, achieving a comprehensive abstraction of the battery power system and obtaining a multi-scale imaging model of the battery power system, including:

[0066] Cell-to-module scale modeling: At the cell scale, an electrochemical-thermal model is established based on the P2D model according to the energy conservation equation, resulting in the cell-scale electrochemical-thermal model; at the single-cell scale, a power battery thermo-mechanical coupling model is constructed based on the thermal stress and thermal expansion behavior caused by the temperature gradient during battery operation, resulting in the single-cell-scale thermo-mechanical coupling model.

[0067] Modeling at the module and overall scale: Based on the electrochemical-thermal model at the cell scale and the thermo-mechanical coupling model at the single cell scale, a three-dimensional CAD model of the power battery is constructed using industrial modeling software to accurately express its key structural parameters, constraints and positioning relationships between single cells and modules, and obtain a three-dimensional CAD model; and geometric optimization is performed using PixyzStudio to obtain an optimized geometric model;

[0068] Dynamic modeling: Using the optimized geometric model as input, and combining the mechanical geometry, material parameters, force analysis, and velocity load information of the power battery pack, motor, inverter, and drive control unit, a dynamic model containing mechanical geometry, material parameters, force analysis, and velocity load information is established.

[0069] In this embodiment, based on the acquired basic data of the new energy heavy-duty truck battery power system, models are constructed from different scales and levels to gradually achieve a comprehensive abstraction of the battery power system, thereby obtaining an accurate multi-scale imaging model. Traditional battery power system modeling often focuses only on a single scale or a single physical field, making it difficult to fully reflect the complex characteristics of the system. This solution, however, considers the coupling effects of multiple physical fields such as electrochemistry, heat, and force, as well as the dynamic characteristics of the system, from the cell to the module, single cell, module, and entire pack scales, organically combining models at different scales. Through this hierarchical, multi-dimensional modeling approach, the behavior and performance of the battery power system under different operating conditions can be accurately captured, providing a solid foundation for the subsequent construction of a digital twin model. This significantly improves the accuracy of predicting and analyzing the state of the battery power system, and helps optimize system design and operation strategies.

[0070] Specifically, the implementation process of this embodiment is as follows:

[0071] The process of modeling from cell to module scale includes:

[0072] Cell-scale electrochemical-thermal model: At the cell scale, an electrochemical-thermal model is established based on the P2D (pseudo-two-dimensional) model, according to the energy conservation equation. The P2D model is a commonly used model to describe the internal electrochemical processes of lithium-ion batteries. The energy conservation equation can be expressed as:

[0073]

[0074] Where ρ is the density of the cell material, and c p It is specific heat capacity, T is temperature, k is thermal conductivity, and Q is... rxn It is the heat production rate of a chemical reaction, Q. ohm This refers to the ohmic heat generation rate. By solving this equation and combining it with the descriptions of electrochemical processes such as electrode reactions and ion transport in the P2D model, an electrochemical-thermal model at the cell scale is obtained. Cell-scale thermo-mechanical coupling model: At the cell scale, the thermal stress and thermal expansion behavior caused by the temperature gradient during battery operation are considered. Thermal stress can be calculated based on thermoelasticity theory, and thermal expansion behavior can be described by the coefficient of thermal expansion. Assuming the battery material is isotropic, the thermal stress σ... th It can be represented as:

[0075] σ th =EαΔT

[0076] Where E is the elastic modulus of the material, α is the coefficient of thermal expansion, and ΔT is the temperature change. Based on this, a thermo-mechanical coupling model of the power battery is constructed by combining the mechanical equilibrium equation and the heat conduction equation, resulting in a thermo-mechanical coupling model at the cell scale.

[0077] Module and whole-package scale modeling, the process includes:

[0078] 3D CAD Model Construction: Based on the electrochemical-thermal model at the cell scale and the thermo-mechanical coupling model at the single-cell scale, a 3D CAD model of the power battery is constructed using industrial modeling software (such as CATIA, SolidWorks, etc.). During the modeling process, key structural parameters (such as cell size, module layout, etc.) and the constraints and positioning relationships between single cells and modules are accurately expressed. Through accurate modeling of these parameters and relationships, the physical structure of the battery power system can be realistically reflected.

[0079] Geometric Optimization: Pixyz Studio is used to perform geometric optimization on the constructed 3D CAD model. This software can perform operations such as surface subdivision, topology correction, extraction, and repair. For example, surface subdivision can improve the surface accuracy of the model, and topology correction can correct topological errors in the model, thereby obtaining an optimized geometric model and improving the quality and analyzability of the model.

[0080] Dynamic modeling, the process includes:

[0081] The optimized geometric model is used as input, combined with the mechanical geometry, material parameters, force analysis, and velocity-load information of the power battery pack, motor, inverter, and drive control unit. Based on Newton's second law and principles of dynamics, dynamic equations are established. Assume the total mass of the system is m, and the total external force is F. total If the acceleration is a, then the dynamic equation can be expressed as:

[0082] F total =m·a

[0083] Simultaneously, considering factors such as the motor torque and the inverter power conversion, corresponding mathematical models are established to describe the interaction between these components. For example, the motor torque T can be expressed as:

[0084] T = k·I

[0085] Where k is the torque constant of the motor, and I is the current. By comprehensively considering these factors, a dynamic model is established that includes information on mechanical geometry, material parameters, force analysis, and velocity load.

[0086] As one embodiment, the step of adding a data-driven model, extracting information from the real-time monitoring data of the battery power system, and supplementing the extracted information into the multi-scale imaging model to construct a digital twin model of the battery power system includes:

[0087] Model system fusion: The multi-scale image model obtained in the multi-scale modeling step is fused with the data-driven model to form the model system before fusion;

[0088] Real-time data processing: A data-driven approach is used to extract information from real-time monitoring data. The extracted information is then added to the multi-scale imaging model in the pre-fusion model system to form a supplemented multi-scale imaging model. The real-time monitoring data includes sensor data from the physical layer and simulation data from the virtual layer.

[0089] Digital twin construction: Based on the supplemented multi-scale imaging model, a multi-scale digital twin model of the battery power system is constructed to achieve data-driven model fusion.

[0090] In this embodiment, a digital twin model that accurately reflects the actual operating state of the battery power system is constructed by integrating a multi-scale imaging model with a data-driven model and updating and improving the model using real-time monitoring data. Traditional battery power system modeling often lacks real-time data updates and dynamic adjustments, making it difficult to adapt to complex and changing actual operating conditions. This solution innovatively introduces a data-driven model, using real-time monitoring data (including sensor data from the physical layer and simulation data from the virtual layer) to supplement and correct the multi-scale imaging model. This approach enables the digital twin model to track the state changes of the battery power system in real time, improving the model's accuracy and reliability, and thus providing stronger support for the optimized control, fault diagnosis, and predictive maintenance of the battery power system in new energy heavy-duty trucks. In summary, multi-scale imaging models suffer from parameter deviations under complex operating conditions, while data-driven models can extract features from real-time data and supplement them into the physical model. This solution, through a two-layer architecture of "multi-scale imaging model + data-driven model," achieves dynamic improvement in model accuracy, enabling the digital twin model to accurately reflect the real-time state of the battery power system, ultimately improving the energy utilization efficiency of new energy heavy-duty trucks.

[0091] Specifically, the implementation process of this embodiment is as follows:

[0092] The model system integration process includes:

[0093] Multiscale imaging model M ms With data-driven model M dd Integration can be achieved in the following ways: in This indicates a feature concatenation operation. For data-driven models (such as LSTM networks), D real-time For real-time monitoring data.

[0094] Real-time data processing, the process includes:

[0095] Feature extraction: for real-time monitoring data D real-time (including sensor data D) sensor and simulation data D sim Apply the following nonnegative matrix factorization (NMF):

[0096]

[0097] in, For complete modal data, For incomplete modal data, U (υ) Let P be the basis matrix. C and To share the feature matrix.

[0098] Additional information: The extracted features P C and Input a multi-scale imaging model, and update the parameters through the following data-driven model complementation mechanism: Where α is the learning rate, ⊙ represents element-wise multiplication, and the ReLU activation function ensures non-negativity.

[0099] The process of building a digital twin includes:

[0100] Based on the supplemented multi-scale imaging model M′ ms Build a digital twin model M DT :

[0101] in This is a GCN-LSTM network. The specific implementation is as follows:

[0102] Graph Neural Network Layers: Using GraphSAGE aggregation formulas:

[0103] in Let υ be the set of the k-th layer neighbor nodes of node υ.

[0104] Long Short-Term Memory (LSTM) network layers: The following LSTM units are applied to the spatiotemporal features:

[0105] i t =σ(W xi x t +W hi h t-1 +b i ),

[0106] f t =σ(W xf x t +W hf h t-1 +b f ),

[0107] o t =σ(W xo x t +W ho h t-1 +b o ),

[0108]

[0109] h t =o t ☉tanh(c t ),

[0110] Where x t h is the current input. t To be in a hidden state, c t These are memory cells.

[0111] The above process illustrates that the output M of the model system fusion pre-fusion As input for real-time data processing, feature P is extracted using NMF. C and The result of real-time data processing M′ ms As the initial state for constructing a digital twin, spatiotemporal features are fused through a GCN-LSTM network, ultimately outputting M. DT .

[0112] As one embodiment, the method of abstracting the sensor data of the battery power system into a network topology graph using a graph neural network-long short-term memory neural network structure for multi-scale modeling of spatiotemporal relationships, and using the constructed digital twin model of the battery power system to perceive the operating status information of the battery power system, includes:

[0113] Data abstraction processing: The sensor data is abstracted into a network topology diagram of the sensor data time-series accumulation process, and the connections between nodes are defined as sequential events, forming a network topology diagram of the sensor data;

[0114] Dynamic graph construction: Based on the network topology graph of the sensor data, a time-series dynamic graph structure for fusing battery-motor-braking characteristic information is constructed to obtain a time-series dynamic graph containing battery-motor-braking characteristics;

[0115] Feature learning processing: The structural and temporal information of the time-series dynamic graph is integrated using a graph neural network-long short-term memory neural network structure. A hierarchical feature aggregation method is designed to learn feature aggregators in neighborhoods at different depths, so as to realize the flow of information from higher depths to nodes and obtain state perception results.

[0116] In this embodiment, spatiotemporal feature fusion and dynamic graph modeling address the problem that traditional sensor data processing methods cannot effectively capture the multi-physics coupling dynamic characteristics of battery power systems. Sensor data exhibits strong correlation and temporal sequence, but traditional methods only perform single-time-series analysis, leading to feature loss. This solution abstracts sensor data into a network topology graph, constructs a time-series dynamic graph combining battery-motor-braking characteristics, and fuses spatial structure information with time-series information using a GCN-LSTM structure. This solution overcomes the spatiotemporal decoupling limitations of traditional methods, achieving accurate perception of the battery power system's operating state and providing real-time and comprehensive state data for subsequent control decisions.

[0117] Specifically, the implementation process of this embodiment is as follows:

[0118] Data abstraction and processing includes:

[0119] Network topology construction: Abstract the sensor data into a set G = (V, E, X, L), where: V = {υ1, υ2, ..., υ...} n} represents the set of sensor nodes (such as voltage, temperature, and current sensors); E = {(υ i υ j ) | Node υ i With υ j There is a temporal correlation. For node features (such as sensor measurements); L = {l1, l2, ..., l...} m} represents node labels (such as battery status categories).

[0120] Spatiotemporal relationship modeling: using adjacency matrix Indicates node connection, where A i，j =1 if and only if (υ i υ j )∈E.

[0121] Dynamic graph construction, the process includes:

[0122] Feature information fusion: Battery characteristics (such as SOC, internal resistance), motor characteristics (such as torque, speed), and braking characteristics (such as braking force distribution) are mapped to network nodes to form a multimodal feature matrix.

[0123] Time series dynamic graph generation: Using the GraphSAGE method, time series data can be transformed into dynamic graphs. Among them G (t) This represents the dynamic graph at time t, where AGGR is the time series aggregation function.

[0124] Feature learning processing, the process includes:

[0125] GCN-LSTM structure:

[0126] Graph convolutional layer: using aggregation formula: in Let υ be the set of the k-th level neighbors. Let be the embedding vector of the k-th layer.

[0127] LSTM layer: Apply LSTM units to the embedding vectors of each layer:

[0128] i t =σ(W xi x t +W hi h t-1 +b i ),

[0129] f t =σ(W xf x t +W hf h t-1 +b f ),

[0130] o t =σ(W xo x t +W ho h t-1 +b o ),

[0131]

[0132] h t =o t ☉tanh(c t ),

[0133] Where x t For embedding the current graph, h t To be in a hidden state, c t These are memory cells.

[0134] Hierarchical Feature Aggregation: A hierarchical feature aggregation method is designed to learn feature aggregators for each of the k-level neighborhoods. Where K is the maximum neighborhood depth, and CONCAT represents feature concatenation.

[0135] Supervised learning training: Training the model using the classification cross-entropy loss function. Where yυ，c Let node υ be the true label of category c. To predict probabilities.

[0136] The above process illustrates that the output G of data abstraction processing serves as the input for dynamic graph construction, and G is generated through the fusion of characteristic information. (t) The result of dynamic graph construction G (t) The input features are processed through learning, and spatiotemporal features are extracted using GCN-LSTM to finally output the state-aware results.

[0137] As one embodiment, the method of utilizing a cross-modal deep semantic matching mechanism to construct a shared feature subspace among modalities for the multimodal data contained in the perceived operational state information, and designing an invariant graph regularization factor to establish a deep semantic matching fusion model for incomplete multimodal and multi-physics fields, thereby reducing semantic bias, includes:

[0138] Cross-modal modeling: A cross-modal deep semantic matching mechanism is adopted, and an incomplete cross-modal deep semantic matching fusion model is established through the multi-level and multi-scale nonlinear correlation between modal data to obtain the initial fusion model;

[0139] Feature space construction: Based on the initial fusion model, a shared feature subspace between modalities is constructed, and the sharing of incomplete cross-modal data is learned to form a shared feature subspace;

[0140] Regularization process: Invariant graph regularization factors are designed in the shared feature subspace to ensure the local similarity characteristics of each modality data, and the regularized shared subspace is obtained;

[0141] Objective function establishment: A new objective function is established based on the regularized shared subspace to form a complete multimodal data fusion model, thereby reducing semantic bias.

[0142] In this embodiment, a cross-modal deep semantic matching mechanism is used to address the semantic bias problem in the fusion of incomplete multimodal and multiphysics data for multimodal data, targeting the multimodal data in the operating status information of new energy heavy-duty truck battery power systems. Traditional data fusion methods struggle to handle complex nonlinear relationships and missing data among multimodal data, leading to inaccurate fusion results. This solution first establishes an initial fusion model through a cross-modal deep semantic matching mechanism to capture multi-level and multi-scale nonlinear correlations between modal data; then, it constructs a shared feature subspace between modalities to learn the sharing of incomplete cross-modal data; next, it designs an invariant graph regularization factor to ensure the local similarity characteristics of each modal data; finally, it establishes a new objective function to form a complete multimodal data fusion model. This approach effectively reduces semantic bias, improves the accuracy of multimodal data fusion, and provides more reliable data support for the precise analysis and decision-making of battery power systems.

[0143] Specifically, the implementation process of this embodiment is as follows:

[0144] Cross-modal modeling: A cross-modal deep semantic matching mechanism is employed to establish an incomplete cross-modal deep semantic matching fusion model. Assume the multimodal data has M modalities, and the data of the m-th modality is represented by X. m m = 1, 2, ..., M. Through multi-level nonlinear transformation function f m Process the modal data as follows:

[0145] H m =f m (X m )

[0146] Where H m This is the characteristic representation of the m-th mode after transformation. Then, considering the multi-level nonlinear correlation between modes, an initial fusion model F0 is established, which can be expressed as: F0 = g(H1, H2, ..., H...). M ),

[0147] Here, g is a nonlinear function that integrates features from various modalities, such as a multilayer perceptron (MLP) structure.

[0148] Feature Space Construction: A shared feature subspace among modes is constructed based on the initial fusion model F0. The Non-negative Matrix Factorization (NMF) method is used. Assuming the basis matrix of the shared feature subspace is U and the coefficient matrix is ​​P, for each mode, the transformed feature H... m ,have:

[0149]

[0150] stU≥0, P m ≥0

[0151] By solving the above optimization problem, we obtain the shared basis matrix U and the coefficient matrix P corresponding to each mode. m This creates a shared feature subspace.

[0152] Regularization: Invariant graph regularization factors are designed in the shared feature subspace. An invariant graph regularization factor R is defined to guarantee the local similarity characteristics of each modality's data. Let S be the local similarity matrix of each modality's data, which can be obtained, for example, by calculating the Euclidean distance or cosine similarity between data points. Then the invariant graph regularization factor R can be expressed as:

[0153]

[0154] Among them U i and U jThese are the i-th and j-th rows of the shared basis matrix U. Adding a regularization factor R to the original objective function yields the regularized objective function:

[0155]

[0156] Here, λ is the regularization coefficient, and its value is adjusted to balance the constraints of data fitting and local similarity. Solving the above regularized objective function yields the regularized shared basis matrix U′ and coefficient matrix P′. m This results in a regularized shared subspace.

[0157] Objective function establishment: A new objective function is established based on the regularized shared subspace. Let the regularized shared basis matrix be U′ and the coefficient matrix be P′. m For prediction tasks (such as battery state classification), let the true label be Y and the predicted label be Y. The cross-entropy loss function (suitable for classification problems) or the mean squared error loss function (suitable for regression problems) can be used. Taking the cross-entropy loss function as an example, the new objective function L can be expressed as:

[0158]

[0159] Where Y i，c It is the true label of the i-th sample belonging to the c-th class. It predicts probabilities. By minimizing the objective function L and continuously adjusting the model parameters, a complete multimodal data fusion model is formed, thereby reducing semantic bias and improving the model's accuracy in predicting the operating state of the battery power system.

[0160] As one example, once a deep semantic matching and fusion model of an incomplete multimodal and multiphysical field is established, this model becomes a core enabling technology for the digital twin of the battery power system of new energy heavy-duty trucks. By eliminating the semantic gap between multi-source data, it achieves high-precision mapping between physical entities and virtual models, ultimately supporting real-time optimization control and full life-cycle management of the battery power system. Furthermore, this model can be applied in the following ways:

[0161] Real-time status monitoring and prediction: By integrating multimodal data of the battery power system (such as electrochemical parameters, temperature distribution, mechanical stress, etc.), this model can analyze the comprehensive performance of the battery under complex operating conditions in real time. For example, by combining cell-level SOC (state of charge), module-level temperature gradient, and pack-level stress distribution, it can predict the available capacity degradation trend of the battery pack under different driving conditions.

[0162] Energy management optimization: Based on a deep correlation analysis of battery-motor-braking characteristics using a fusion model, regenerative braking energy recovery strategies can be optimized. For example, the braking force distribution ratio can be dynamically adjusted according to real-time battery charging capacity constraints and motor power generation characteristic constraints to maximize energy recovery rate.

[0163] Fault diagnosis and lifespan assessment: By utilizing multiphysics coupling features extracted from a shared feature subspace, potential fault modes within the battery can be identified. For example, by analyzing the spatiotemporal correlation characteristics of temperature and stress fields, the development trend of microcracks within the cell can be located, enabling accurate prediction of battery lifespan.

[0164] Understandably, this fusion model serves as the "semantic engine" of the digital twin system, providing the virtual model with a deep understanding of multimodal data. It semantically aligns physical layer sensor data with virtual layer simulation data through a shared feature subspace, ensuring the digital twin model can reflect the real-time state of the battery power system. The output of the fusion model (such as state perception results) acts as a feedback signal, continuously calibrating the parameters of the digital twin model. For example, when the temperature difference between battery cells exceeds a threshold, the thermal management strategy optimization of the digital twin model is automatically triggered. The semantic matching results of the model are directly input into the online optimal control mechanism driven by the digital twin. For instance, in a multi-agent collaborative control framework, the battery health status assessment results provided by the fusion model can guide the energy management agent to dynamically adjust the charging and discharging strategies.

[0165] See Figure 2 This is a schematic diagram of the structure of a digital twin system for a battery power system applied to a new energy heavy-duty truck, according to an embodiment of the present invention. The digital twin system for the battery power system applied to a new energy heavy-duty truck includes:

[0166] The acquisition module 10 is used to acquire the basic data of the digital twin of the battery power system applied to the new energy heavy truck. The basic data includes: full-element parameter information of the battery power system of the new energy heavy truck under the multi-physics field coupling state of electrochemical field, temperature field and stress field, as well as multi-scale data from battery cell to module, pack, controller and drive motor.

[0167] The first construction module 11 is used to construct an electrochemical-thermal coupling model from the cell to the module scale of the battery power system based on the basic data, construct a thermo-mechanical coupling model at the cell scale considering thermal stress and thermal expansion behavior at the cell scale, establish and optimize a geometric model from the module and pack scales of the battery power system, and simultaneously establish a dynamic model of the battery power system to achieve a comprehensive abstraction of the battery power system and obtain a multi-scale image model of the battery power system.

[0168] The second construction module 12 is used to add a data-driven model, extract information from the real-time monitoring data of the battery power system, supplement the extracted information into the multi-scale image model, and construct a digital twin model of the battery power system.

[0169] The third construction module 13 is used to abstract the sensor data of the battery power system into a network topology graph through a graph neural network-long short-term memory neural network structure to perform multi-scale modeling of spatiotemporal relationships, and to use the constructed digital twin model of the battery power system to perceive the operating status information of the battery power system.

[0170] The fourth construction module 14 is used to utilize a cross-modal deep semantic matching mechanism to construct a shared feature subspace between modalities for the multimodal data contained in the perceived operating state information, and to design an invariant graph regularization factor to establish a deep semantic matching fusion model for incomplete multimodal multiphysics fields, thereby reducing semantic bias.

[0171] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0172] This invention constructs a multi-physics coupled, multi-scale digital twin model, combines a data-driven approach to achieve dynamic model updates, and utilizes spatiotemporal feature fusion and cross-modal semantic matching techniques to process multi-source heterogeneous data. Specifically, it first collects all-element parameters and multi-scale data of the multi-physics field (electrochemical field, temperature field, stress field) of the battery power system to construct a multi-level physical model encompassing the entire battery pack from the cell level. Then, it supplements the model with real-time monitoring data through a data-driven model, improving model accuracy. Next, it employs a graph neural network-long short-term memory neural network structure to process the spatiotemporal relationships of sensor data, achieving precise perception of operational status. Finally, it establishes a shared feature subspace and introduces a regularization factor through a cross-modal semantic matching mechanism to address the problem of incomplete multimodal data. This technical concept, through a hierarchical model construction and data fusion method, overcomes the limitations of traditional models that rely solely on apparent parameters and lack dynamic adaptability, achieving comprehensive, high-precision simulation and real-time optimization of the battery power system. This invention addresses the technical shortcomings of traditional methods that cannot dynamically reflect the complex operating conditions of batteries by using multi-physics coupling modeling and real-time data-driven mechanisms. It achieves accurate mapping and real-time optimization of the multi-scale operating states of the power system of new energy heavy truck batteries, significantly improving the system's energy utilization efficiency and dynamic response capability.

[0173] As one embodiment, the first construction module is specifically used for:

[0174] Cell-to-module scale modeling: At the cell scale, an electrochemical-thermal model is established based on the P2D model according to the energy conservation equation, resulting in the cell-scale electrochemical-thermal model; at the single-cell scale, a power battery thermo-mechanical coupling model is constructed based on the thermal stress and thermal expansion behavior caused by the temperature gradient during battery operation, resulting in the single-cell-scale thermo-mechanical coupling model.

[0175] Modeling at the module and overall scale: Based on the electrochemical-thermal model at the cell scale and the thermo-mechanical coupling model at the single cell scale, a three-dimensional CAD model of the power battery is constructed using industrial modeling software to accurately express its key structural parameters, constraints and positioning relationships between single cells and modules, and obtain a three-dimensional CAD model; and geometric optimization is performed using PixyzStudio to obtain an optimized geometric model;

[0176] Dynamic modeling: Using the optimized geometric model as input, and combining the mechanical geometry, material parameters, force analysis, and velocity load information of the power battery pack, motor, inverter, and drive control unit, a dynamic model containing mechanical geometry, material parameters, force analysis, and velocity load information is established.

[0177] As one embodiment, the second building module is specifically used for:

[0178] Model system fusion: The multi-scale image model obtained in the multi-scale modeling step is fused with the data-driven model to form the model system before fusion;

[0179] Real-time data processing: A data-driven approach is used to extract information from real-time monitoring data. The extracted information is then added to the multi-scale imaging model in the pre-fusion model system to form a supplemented multi-scale imaging model. The real-time monitoring data includes sensor data from the physical layer and simulation data from the virtual layer.

[0180] Digital twin construction: Based on the supplemented multi-scale imaging model, a multi-scale digital twin model of the battery power system is constructed to achieve data-driven model fusion.

[0181] As one embodiment, the third building module is specifically used for:

[0182] Data abstraction processing: The sensor data is abstracted into a network topology diagram of the sensor data time-series accumulation process, and the connections between nodes are defined as sequential events, forming a network topology diagram of the sensor data;

[0183] Dynamic graph construction: Based on the network topology graph of the sensor data, a time-series dynamic graph structure for fusing battery-motor-braking characteristic information is constructed to obtain a time-series dynamic graph containing battery-motor-braking characteristics;

[0184] Feature learning processing: The structural and temporal information of the time-series dynamic graph is integrated using a graph neural network-long short-term memory neural network structure. A hierarchical feature aggregation method is designed to learn feature aggregators in neighborhoods at different depths, so as to realize the flow of information from higher depths to nodes and obtain state perception results.

[0185] As one embodiment, the fourth building module is specifically used for:

[0186] Cross-modal modeling: A cross-modal deep semantic matching mechanism is adopted, and an incomplete cross-modal deep semantic matching fusion model is established through the multi-level and multi-scale nonlinear correlation between modal data to obtain the initial fusion model;

[0187] Feature space construction: Based on the initial fusion model, a shared feature subspace between modalities is constructed, and the sharing of incomplete cross-modal data is learned to form a shared feature subspace;

[0188] Regularization process: Invariant graph regularization factors are designed in the shared feature subspace to ensure the local similarity characteristics of each modality data, and the regularized shared subspace is obtained;

[0189] Objective function establishment: A new objective function is established based on the regularized shared subspace to form a complete multimodal data fusion model, thereby reducing semantic bias.

[0190] It should be noted that the system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, in the accompanying drawings of the system embodiments provided by this invention, the connection relationships between modules indicate that they have communication connections, which can be specifically implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement this without any creative effort.

[0191] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A digital twin method for battery power systems applied to new energy heavy-duty trucks, characterized in that, Includes the following steps: Acquire basic data for the digital twin of the battery power system applied to new energy heavy trucks. The basic data includes: full-element parameter information of the battery power system of new energy heavy trucks under the multi-physics field coupling state of electrochemical field, temperature field and stress field, as well as multi-scale data from battery cells to modules, packs, controllers and drive motors. Based on the aforementioned fundamental data, an electrochemical-thermal coupling model is constructed from the cell to the module scale of the battery power system. A thermo-mechanical coupling model is constructed at the cell scale considering thermal stress and thermal expansion behavior. A geometric model is established and optimized from the module and pack scales of the battery power system. At the same time, a dynamic model of the battery power system is established to achieve a comprehensive abstraction of the battery power system and obtain a multi-scale image model of the battery power system. A data-driven model is incorporated to extract information from the real-time monitoring data of the battery power system. The extracted information is then added to the multi-scale imaging model to construct a digital twin model of the battery power system. By using a graph neural network-long short-term memory neural network structure, the sensor data of the battery power system is abstracted into a network topology graph for multi-scale modeling of spatiotemporal relationships. The constructed digital twin model of the battery power system is then used to perceive the operating status information of the battery power system. By utilizing a cross-modal deep semantic matching mechanism, a shared feature subspace between modalities is constructed for the multimodal data contained in the perceived operational state information, and an invariant graph regularization factor is designed to establish a deep semantic matching fusion model for incomplete multimodal and multi-physics fields, thereby reducing semantic bias. The step of adding a data-driven model involves extracting information from the real-time monitoring data of the battery power system, supplementing the extracted information into the multi-scale imaging model, and constructing a digital twin model of the battery power system, including: Model system fusion: The multi-scale mapping model obtained in the multi-scale modeling steps is fused with the data-driven model to form the model system before fusion; among them, the multi-scale mapping model With data-driven models Integration can be achieved in the following ways: , where ⊕ represents the feature concatenation operation, For data-driven models, For real-time monitoring data; Real-time data processing: A data-driven approach is used to extract information from real-time monitoring data. This extracted information is then added to the multi-scale imaging model within the pre-fusion model system, forming a supplemented multi-scale imaging model. The real-time monitoring data includes sensor data from the physical layer and simulation data from the virtual layer. Real-time data processing includes: processing the real-time monitoring data... Apply the following nonnegative matrix factorization: ,in, For complete modal data, This is incomplete modal data. As a basis matrix, and To share the feature matrix; the extracted features and Input a multi-scale imaging model, and update the parameters through the following data-driven model complementation mechanism: , in Here, ⊙ represents the learning rate, and ReLU activation function ensures non-negativity. Digital twin construction: Based on the supplemented multi-scale imaging model, a multi-scale digital twin model of the battery power system is constructed to achieve data-driven model fusion.

2. The digital twin method for battery power systems applied to new energy heavy trucks according to claim 1, characterized in that, Based on the aforementioned fundamental data, an electrochemical-thermal coupling model is constructed from the cell to the module scale of the battery power system. A thermo-mechanical coupling model is constructed at the cell scale, considering thermal stress and thermal expansion behavior. Geometric models are established and optimized at the module and overall pack scales of the battery power system. Simultaneously, a dynamic model of the battery power system is established, achieving a comprehensive abstraction of the battery power system and obtaining a multi-scale imaging model of the battery power system, including: Cell-to-module scale modeling: At the cell scale, an electrochemical-thermal model is established based on the P2D model according to the energy conservation equation, resulting in the cell-scale electrochemical-thermal model; at the single-cell scale, a power battery thermo-mechanical coupling model is constructed based on the thermal stress and thermal expansion behavior caused by the temperature gradient during battery operation, resulting in the single-cell-scale thermo-mechanical coupling model. Modeling at the module and overall scale: Based on the electrochemical-thermal model at the cell scale and the thermo-mechanical coupling model at the single cell scale, a three-dimensional CAD model of the power battery is constructed using industrial modeling software to accurately express its key structural parameters, constraints and positioning relationships between single cells and modules, and obtain a three-dimensional CAD model; and geometric optimization is performed using PixyzStudio to obtain an optimized geometric model; Dynamic modeling: Using the optimized geometric model as input, and combining the mechanical geometry, material parameters, force analysis, and velocity load information of the power battery pack, motor, inverter, and drive control unit, a dynamic model containing mechanical geometry, material parameters, force analysis, and velocity load information is established.

3. The digital twin method for battery power systems applied to new energy heavy trucks according to claim 1, characterized in that, The method involves using a graph neural network-long short-term memory neural network structure to abstract the sensor data of the battery power system into a network topology graph for multi-scale modeling of spatiotemporal relationships. The constructed digital twin model of the battery power system is then used to perceive the operating status information of the battery power system, including: Data abstraction processing: The sensor data is abstracted into a network topology diagram of the sensor data time-series accumulation process, and the connections between nodes are defined as sequential events, forming a network topology diagram of the sensor data; Dynamic graph construction: Based on the network topology graph of the sensor data, a time-series dynamic graph structure for fusing battery-motor-braking characteristic information is constructed to obtain a time-series dynamic graph containing battery-motor-braking characteristics; Feature learning processing: The structural and temporal information of the time-series dynamic graph is integrated using a graph neural network-long short-term memory neural network structure. A hierarchical feature aggregation method is designed to learn feature aggregators in neighborhoods at different depths, so as to realize the flow of information from higher depths to nodes and obtain state perception results.

4. The digital twin method for battery power systems applied to new energy heavy trucks according to claim 1, characterized in that, The method utilizes a cross-modal deep semantic matching mechanism to construct a shared feature subspace among modalities for the multimodal data contained in the perceived operational state information, and designs an invariant graph regularization factor to establish a deep semantic matching fusion model for incomplete multimodal and multi-physics fields, thereby reducing semantic bias. This includes: Cross-modal modeling: A cross-modal deep semantic matching mechanism is adopted, and an incomplete cross-modal deep semantic matching fusion model is established through the multi-level and multi-scale nonlinear correlation between modal data to obtain the initial fusion model; Feature space construction: Based on the initial fusion model, a shared feature subspace between modalities is constructed, and the sharing of incomplete cross-modal data is learned to form a shared feature subspace; Regularization process: Invariant graph regularization factors are designed in the shared feature subspace to ensure the local similarity characteristics of each modality data, and the regularized shared subspace is obtained; Objective function establishment: A new objective function is established based on the regularized shared subspace to form a complete multimodal data fusion model, thereby reducing semantic bias.

5. A digital twin system for battery power systems applied to new energy heavy-duty trucks, characterized in that, include: The acquisition module is used to acquire the basic data of the digital twin of the battery power system applied to new energy heavy trucks. The basic data includes: full-element parameter information of the battery power system of new energy heavy trucks under the multi-physics field coupling state of electrochemical field, temperature field and stress field, as well as multi-scale data from battery cells to modules, packs, controllers and drive motors. The first construction module is used to construct an electrochemical-thermal coupling model of the battery power system from the cell to the module scale based on the basic data, construct a thermo-mechanical coupling model at the cell scale considering thermal stress and thermal expansion behavior at the cell scale, establish and optimize a geometric model from the module and pack scales of the battery power system, and establish a dynamic model of the battery power system at the same time, so as to realize a comprehensive abstraction of the battery power system and obtain a multi-scale image model of the battery power system. The second construction module is used to add a data-driven model, extract information from the real-time monitoring data of the battery power system, supplement the extracted information into the multi-scale imaging model, and construct a digital twin model of the battery power system. The third construction module is used to abstract the sensor data of the battery power system into a network topology graph through a graph neural network-long short-term memory neural network structure to perform multi-scale modeling of spatiotemporal relationships, and to use the constructed digital twin model of the battery power system to perceive the operating status information of the battery power system. The fourth construction module is used to utilize a cross-modal deep semantic matching mechanism to construct a shared feature subspace between modalities for the multimodal data contained in the perceived operational state information, and to design an invariant graph regularization factor to establish a deep semantic matching fusion model for incomplete multimodal and multi-physics fields, thereby reducing semantic bias. The second building module is specifically used for: Model system fusion: The multi-scale mapping model obtained in the multi-scale modeling steps is fused with the data-driven model to form the model system before fusion; among them, the multi-scale mapping model With data-driven models Integration can be achieved in the following ways: , where ⊕ represents the feature concatenation operation, For data-driven models, For real-time monitoring data; Real-time data processing: A data-driven approach is used to extract information from real-time monitoring data. This extracted information is then added to the multi-scale imaging model within the pre-fusion model system, forming a supplemented multi-scale imaging model. The real-time monitoring data includes sensor data from the physical layer and simulation data from the virtual layer. Real-time data processing includes: processing the real-time monitoring data... Apply the following nonnegative matrix factorization: ,in, For complete modal data, This is incomplete modal data. As a basis matrix, and To share the feature matrix; the extracted features and Input a multi-scale imaging model, and update the parameters through the following data-driven model complementation mechanism: , in Here, ⊙ represents the learning rate, and ReLU activation function ensures non-negativity. Digital twin construction: Based on the supplemented multi-scale imaging model, a multi-scale digital twin model of the battery power system is constructed to achieve data-driven model fusion.

6. The digital twin system for battery power systems applied to new energy heavy-duty trucks as described in claim 5, characterized in that, The first building module is specifically used for: Cell-to-module scale modeling: At the cell scale, an electrochemical-thermal model is established based on the P2D model according to the energy conservation equation, resulting in the cell-scale electrochemical-thermal model; at the single-cell scale, a power battery thermo-mechanical coupling model is constructed based on the thermal stress and thermal expansion behavior caused by the temperature gradient during battery operation, resulting in the single-cell-scale thermo-mechanical coupling model. Modeling at the module and overall scale: Based on the electrochemical-thermal model at the cell scale and the thermo-mechanical coupling model at the single cell scale, a three-dimensional CAD model of the power battery is constructed using industrial modeling software to accurately express its key structural parameters, constraints and positioning relationships between single cells and modules, and obtain a three-dimensional CAD model; and geometric optimization is performed using PixyzStudio to obtain an optimized geometric model; Dynamic modeling: Using the optimized geometric model as input, and combining the mechanical geometry, material parameters, force analysis, and velocity load information of the power battery pack, motor, inverter, and drive control unit, a dynamic model containing mechanical geometry, material parameters, force analysis, and velocity load information is established.

7. The digital twin system for battery power systems applied to new energy heavy-duty trucks as described in claim 5, characterized in that, The third building module is specifically used for: Data abstraction processing: The sensor data is abstracted into a network topology diagram of the sensor data time-series accumulation process, and the connections between nodes are defined as sequential events, forming a network topology diagram of the sensor data; Dynamic graph construction: Based on the network topology graph of the sensor data, a time-series dynamic graph structure for fusing battery-motor-braking characteristic information is constructed to obtain a time-series dynamic graph containing battery-motor-braking characteristics; Feature learning processing: The structural and temporal information of the time-series dynamic graph is integrated using a graph neural network-long short-term memory neural network structure. A hierarchical feature aggregation method is designed to learn feature aggregators in neighborhoods at different depths, so as to realize the flow of information from higher depths to nodes and obtain state perception results.

8. The digital twin system for battery power systems applied to new energy heavy trucks as described in claim 5, characterized in that, The fourth building module is specifically used for: Cross-modal modeling: A cross-modal deep semantic matching mechanism is adopted, and an incomplete cross-modal deep semantic matching fusion model is established through the multi-level and multi-scale nonlinear correlation between modal data to obtain the initial fusion model; Feature space construction: Based on the initial fusion model, a shared feature subspace between modalities is constructed, and the sharing of incomplete cross-modal data is learned to form a shared feature subspace; Regularization process: Invariant graph regularization factors are designed in the shared feature subspace to ensure the local similarity characteristics of each modality data, and the regularized shared subspace is obtained; Objective function establishment: A new objective function is established based on the regularized shared subspace to form a complete multimodal data fusion model, thereby reducing semantic bias.

Images