Electromagnetic compatibility simulation method and device, electronic equipment, medium and product

Abstract

The invention discloses an electromagnetic compatibility simulation method and device, electronic equipment, a medium and a product. The method comprises the following steps: constructing a target radio wave propagation model based on reinforcement learning; constructing a lightweight dynamic propagation model based on the radio wave propagation model; deploying the lightweight dynamic propagation model to a vehicle-mounted ECU of a to-be-simulated vehicle; generating a target interference source signal according to a pre-constructed interference source model; and injecting the target interference source signal into vehicle-mounted OBU equipment of a to-be-simulated vehicle, and performing electromagnetic compatibility simulation according to a preset traffic scene simulator and a lightweight dynamic propagation model to obtain a simulation result. The method can solve the problems that the change of a complex emergency scene cannot be responded in real time, and the nonlinear characteristic of real vehicle-mounted hardware cannot be considered.

Description

Technical Field

[0001] This application relates to the field of vehicle simulation technology, specifically to an electromagnetic compatibility simulation method, apparatus, electronic device, readable storage medium, and computer program product. Background Technology

[0002] Currently, with the rapid development of autonomous driving technology, the reliability of vehicle-to-everything (V2X) communication has become crucial. The dynamic changes in radio wave propagation characteristics caused by high-speed vehicle movement pose a severe challenge to electromagnetic compatibility (EMC) simulation, demanding extremely high real-time performance and accuracy. Existing methods typically employ offline map interpolation for radio wave propagation simulation. This method relies on pre-generated static environmental data, making it difficult to handle complex and ever-changing sudden traffic scenarios, resulting in significant simulation lag. Furthermore, existing solutions are mostly limited to pure software simulation, failing to consider the nonlinear characteristics of real-world vehicle hardware, leading to discrepancies between simulation results and actual device behavior. Summary of the Invention

[0003] In view of the above problems, this application provides an electromagnetic compatibility simulation method, device, electronic device, readable storage medium and computer program product, which can solve the problems of being unable to respond in real time to complex sudden changes in scenarios and failing to consider the nonlinear characteristics of real vehicle hardware.

[0004] Firstly, this application provides an electromagnetic compatibility simulation method, including: Construct a target radio wave propagation model based on reinforcement learning; A lightweight dynamic propagation model is constructed based on the aforementioned radio wave propagation model; The lightweight dynamic propagation model is deployed on the vehicle ECU of the vehicle to be simulated; Generate target interference source signals based on a pre-built interference source model; The target interference source signal is injected into the on-board unit (OBU) of the vehicle to be simulated, and electromagnetic compatibility simulation is performed according to the preset traffic scenario simulator and the lightweight dynamic propagation model to obtain the simulation results.

[0005] In the above technical solution, this method can effectively get rid of the dependence on pre-generated static environmental data by constructing a target radio wave propagation model based on reinforcement learning and further transforming it into a lightweight dynamic propagation model. This enables real-time response to complex and ever-changing sudden traffic scenarios and solves the problem of simulation lag in existing offline map interpolation methods. At the same time, the lightweight dynamic propagation model is deployed to the on-board ECU of the vehicle to be simulated and the target interference source signal is injected in combination with the real on-board OBU device. This fully incorporates the actual operating characteristics of the on-board hardware, making up for the shortcomings of existing pure software simulations that do not consider the nonlinear characteristics of real on-board hardware. Ultimately, this significantly improves the real-time performance and accuracy of electromagnetic compatibility simulation and greatly reduces the deviation between simulation results and actual equipment behavior.

[0006] In some implementations, constructing a target radio wave propagation model based on reinforcement learning includes: Construct the original radio wave propagation model; Raw signal data is acquired through onboard sensors, onboard communication units, and road test units; Feature extraction is performed on the original signal data to obtain spatiotemporal fusion features; The target radio wave propagation model is obtained by adjusting the model parameters of the original radio wave propagation model using a pre-trained reinforcement learning agent model and the spatiotemporal fusion features.

[0007] In the above technical solution, the method can optimize the parameters of the original radio wave propagation model based on the spatiotemporal fusion features extracted from real vehicle and road test data, combined with a reinforcement learning agent model, to obtain a target radio wave propagation model that is adapted to the actual scenario.

[0008] In some implementations, the step of extracting features from the original signal data to obtain spatiotemporal fusion features includes: Based on the original signal data, obtain building point cloud data and vehicle motion time sequence data; Feature extraction is performed on the building point cloud data to obtain the target spatial features; The vehicle motion time series data is processed by a preset three-layer bidirectional memory network to extract features, obtain the original time series features, and then the original time series features are processed by time synchronization to obtain the target time series features. The target spatial features and the target temporal features are concatenated to obtain concatenated features; The spliced ​​features are adaptively weighted and fused using attention weights to obtain spatiotemporal fusion features.

[0009] In the above technical solution, the method can extract and fuse spatial and temporal features from the original signal data to generate spatiotemporal fusion features with scene adaptability, thereby improving the input quality of the radio wave propagation model.

[0010] In some implementations, the step of extracting features from the building point cloud data to obtain target spatial features includes: Feature extraction is performed on the building point cloud data to obtain spatial structure features; The spatial structure features are normalized and processed using the ReLU activation function to obtain the first processed features; The first processed feature is processed by a max pooling layer to obtain the second processed feature; The second processed feature is evaluated as a vector feature to obtain the target space feature.

[0011] In the above technical solution, the method can extract and process spatial structural features from building point cloud data to obtain high-quality target spatial features suitable for radio wave propagation models.

[0012] In some implementations, the actions performed by the reinforcement learning agent model include path loss exponent, shadow fading variance, and multipath delay scaling factor. The reinforcement learning agent model also includes a multi-objective reward mechanism, which is constructed based on the absolute error between the predicted value and the actual measured value.

[0013] In the above technical solution, the method can use the path loss exponent, shadow fading variance, and multipath delay scaling factor as the action space, and based on the multi-objective reward mechanism of prediction and measurement errors, to achieve accurate optimization and dynamic adjustment of radio wave propagation model parameters.

[0014] In some implementations, constructing a lightweight dynamic propagation model based on the radio wave propagation model includes: A teacher model is constructed in the cloud based on the described radio wave propagation model; Knowledge distillation is performed based on the teacher model to obtain a lightweight student model.

[0015] In the above technical solution, the method can obtain a lightweight student model adapted to the deployment of the vehicle ECU through knowledge distillation of the cloud teacher model, and construct a lightweight dynamic propagation model.

[0016] In some implementations, generating the target interference source signal based on a pre-built interference source model includes: Interference data is synthesized based on a pre-defined interference source database and a pre-built interference source model; wherein the interference source model includes at least a motor broadband noise model and a radar pulse model. An interference source radio frequency signal is generated based on the interference data; A channel simulator was used to simulate a multipath fading environment; The interference source radio frequency signal is subjected to fading processing based on the multipath fading environment to obtain the target interference source signal.

[0017] In the above technical solution, the method can comprehensively consider the characteristics of interference sources such as motor broadband noise and radar pulses to generate target interference source signals that closely resemble the real multipath fading environment.

[0018] In some implementations, the step of performing electromagnetic compatibility simulation based on a preset traffic scenario simulator and the lightweight dynamic propagation model to obtain simulation results includes: The vehicle dynamics are run through a pre-set traffic scenario simulator. Calculate the scene complexity of the vehicle dynamic scene; The target interpolation step size and target calculation method are determined based on the preset dynamic resource scheduling strategy and the complexity of the scenario. Simulation results are obtained by performing simulation calculations based on the target interpolation step size, the target calculation method, the lightweight student model, and the target interference source signal.

[0019] In the above technical solution, the method can combine traffic scene simulator and dynamic scheduling resources based on scene complexity to determine the appropriate interpolation step size and calculation method, and complete electromagnetic compatibility simulation calculation based on lightweight student model and target interference source signal.

[0020] Secondly, this application provides an electromagnetic compatibility simulation device, comprising: The first building unit is used to construct a target radio wave propagation model based on reinforcement learning; The second building unit is used to build a lightweight dynamic propagation model based on the radio wave propagation model. The deployment unit is used to deploy the lightweight dynamic propagation model to the on-board ECU of the vehicle to be simulated; The generation unit is used to generate target interference source signals based on a pre-built interference source model; An interference injection unit is used to inject the target interference source signal into the on-board OBU device of the vehicle to be simulated. The simulation unit is used to perform electromagnetic compatibility simulation based on a preset traffic scenario simulator and the lightweight dynamic propagation model, and obtain simulation results.

[0021] In the above technical solution, the device can effectively get rid of the dependence on pre-generated static environmental data by constructing a target radio wave propagation model based on reinforcement learning and further transforming it into a lightweight dynamic propagation model. This enables it to respond in real time to complex and ever-changing sudden traffic scenarios and solve the problem of simulation lag in existing offline map interpolation methods. At the same time, the lightweight dynamic propagation model is deployed to the on-board ECU of the vehicle to be simulated and the target interference source signal is injected in combination with the real on-board OBU device. This fully incorporates the actual operating characteristics of the on-board hardware, making up for the shortcomings of existing pure software simulations that do not consider the nonlinear characteristics of real on-board hardware. Ultimately, this significantly improves the real-time performance and accuracy of electromagnetic compatibility simulation and greatly reduces the deviation between simulation results and actual equipment behavior.

[0022] Thirdly, this application provides an electronic device, the electronic device including a memory and a processor, the memory for storing a computer program, the processor running the computer program to cause the electronic device to perform the electromagnetic compatibility simulation method described in any one of the first aspects.

[0023] Fourthly, this application provides a readable storage medium storing a computer program, which, when executed by a processor, performs the electromagnetic compatibility simulation method described in any one of the first aspects.

[0024] Fifthly, this application provides a computer program product, which includes a computer program that, when run by a processor, executes the electromagnetic compatibility simulation method described in any one of the first aspects.

[0025] The beneficial effects of this application are: it can accurately simulate the time-varying characteristics of radio wave propagation and dynamically evaluate the quality of wireless communication in the electromagnetic compatibility simulation research of vehicle-to-everything (V2X) networks, providing a theoretical basis for the design optimization and popularization of V2X networks; at the same time, it can also achieve high-precision simulation prediction of V2X communication quality by constructing a mapping mechanism between dynamic traffic scenarios and real-time electromagnetic parameters. Attached Figure Description

[0026] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0027] Figure 1 This is a flowchart illustrating the electromagnetic compatibility simulation method in some embodiments of this application; Figure 2 This is a flowchart illustrating another electromagnetic compatibility simulation method in some embodiments of this application; Figure 3 This is a flowchart illustrating the cloud-vehicle model deployment method in some embodiments of this application; Figure 4 The following is a schematic flowchart illustrating an electromagnetic compatibility simulation method in some embodiments of this application. Figure 5 This is a schematic diagram of the electromagnetic compatibility simulation device in some embodiments of this application; Figure 6 This is a schematic diagram of the structure of an electronic device in some embodiments of this application. Detailed Implementation

[0028] The embodiments of the technical solution of this application will now be described in detail with reference to the accompanying drawings. These embodiments are only used to more clearly illustrate the technical solution of this application and are therefore merely examples, and should not be used to limit the scope of protection of this application.

[0029] Unless otherwise defined, 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 application pertains; the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the application; the terms “comprising” and “having”, and any variations thereof, in the specification, claims, and foregoing description of the drawings are intended to cover non-exclusive inclusion.

[0030] In the description of the embodiments of this application, technical terms such as "first" and "second" are used only to distinguish different objects and should not be construed as indicating or implying relative importance or implicitly specifying the number, specific order, or primary and secondary relationship of the indicated technical features. In the description of the embodiments of this application, "multiple" means two or more (including two), similarly, "multiple sets" refers to two or more sets (including two sets), and "multiple pieces" refers to two or more pieces (including two pieces) unless otherwise explicitly defined.

[0031] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0032] In the description of the embodiments in this application, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this document generally indicates that the preceding and following related objects have an "or" relationship.

[0033] The main challenge in electromagnetic compatibility simulation for vehicle-to-everything (V2X) wireless communication lies in the dynamic changes in radio wave propagation characteristics caused by vehicle motion. Existing "offline map + interpolation" solutions cannot adapt to complex and sudden changes in scenarios.

[0034] To address the aforementioned technical issues, this application provides an electromagnetic compatibility simulation method. This method embeds an online learning module into a step-by-step interpolation algorithm. It collects data such as RSSI (Received Signal Strength Index) and Doppler frequency shift in real time through vehicle-mounted sensors, RSU (Radio Unit), and OBU (On-Board Unit). It uses a reinforcement learning agent (PPO algorithm) to dynamically correct the parameters of the radio wave propagation model based on the collected data. It also uses USRP (United States Repository Programming Language) software radio to reconstruct motor / radar interference and inject it into the actual vehicle-mounted OBU device. Finally, based on the updated propagation model and measured interference data, it performs electromagnetic interference propagation simulation.

[0035] Based on this, the method can overcome the static limitations of offline maps, adopt a reinforcement learning-based dynamic model adjustment to reduce path loss prediction errors and improve model update latency. At the same time, by establishing a dynamic verification system for real equipment in a controllable electromagnetic environment, the method can realize the dynamic verification of real equipment in a controlled electromagnetic environment. Finally, by adopting adaptive multi-precision scheduling to monitor complexity changes in real time, the method can switch simulation methods in a timely manner, effectively reducing the consumption of computing resources.

[0036] like Figure 1 As shown, some embodiments of this application provide an electromagnetic compatibility (EMC) simulation method, which includes: S101. Construct a target radio wave propagation model based on reinforcement learning; S102. Construct a lightweight dynamic propagation model based on the radio wave propagation model; S103. Deploy the lightweight dynamic propagation model onto the on-board ECU of the vehicle to be simulated; S104. Generate the target interference source signal based on the pre-constructed interference source model; S105. Inject the target interference source signal into the on-board OBU device of the vehicle to be simulated, and perform electromagnetic compatibility simulation according to the preset traffic scenario simulator and lightweight dynamic propagation model to obtain the simulation results.

[0037] In some embodiments, the target radio wave propagation model refers to a radio wave propagation model that uses on-board sensors, on-board communication units (communication terminal equipment OBU installed on the vehicle), and roadside units (communication equipment RSU deployed on the side of the road or on traffic facilities) to collect vehicle, communication and environmental data in real time, and dynamically adjusts the model parameters through spatiotemporal feature extraction and reinforcement learning (PPO algorithm) to adapt to vehicle motion characteristics and complex scene changes.

[0038] In some embodiments, the lightweight dynamic propagation model refers to a lightweight, efficient, and deployable radio wave propagation prediction model for an in-vehicle electronic control unit (ECU) obtained through knowledge distillation based on a high-precision teacher model trained in the cloud.

[0039] In some embodiments, an onboard electronic control unit (ECU) refers to an onboard control unit used to carry a lightweight dynamic propagation model, perform real-time radio wave propagation prediction, and participate in electromagnetic compatibility simulation calculations.

[0040] In some embodiments, the interference source model refers to the interference signal model used to generate interference signals that conform to the characteristics of the actual scene, including types such as motor broadband noise and radar pulses.

[0041] In some embodiments, the target interference source signal refers to a real interference signal generated by an interference source model and processed by a channel simulator, which can be injected into the vehicle-mounted OBU device.

[0042] In some embodiments, the on-board unit (OBU) refers to an on-board communication unit that serves as a hardware-in-the-loop test object, receives interference signals, and participates in electromagnetic compatibility simulation.

[0043] In some embodiments, a traffic scene simulator refers to a simulation platform used to run dynamic vehicle scenes, calculate scene complexity, and dynamically schedule simulation resources.

[0044] In some embodiments, the electromagnetic compatibility simulation result refers to the evaluation result of the electromagnetic compatibility of the Internet of Vehicles obtained by combining dynamic scenarios, lightweight propagation models and hardware-in-the-loop interference injection, and through intelligent resource scheduling.

[0045] In the above embodiments, this method can effectively get rid of the dependence on pre-generated static environmental data by constructing a target radio wave propagation model based on reinforcement learning and further transforming it into a lightweight dynamic propagation model. This enables real-time response to complex and ever-changing sudden traffic scenarios and solves the problem of simulation lag in existing offline map interpolation methods. At the same time, by deploying the lightweight dynamic propagation model to the on-board ECU of the vehicle to be simulated and injecting the target interference source signal into the real on-board OBU device, the actual operating characteristics of the on-board hardware are fully integrated, making up for the shortcomings of existing pure software simulations that do not consider the nonlinear characteristics of real on-board hardware. Ultimately, this significantly improves the real-time performance and accuracy of electromagnetic compatibility simulation and greatly reduces the deviation between simulation results and actual equipment behavior.

[0046] In some embodiments, constructing a target radio wave propagation model based on reinforcement learning includes: Construct the original radio wave propagation model; Raw signal data is acquired through onboard sensors, onboard communication units, and road test units; Feature extraction is performed on the original signal data to obtain spatiotemporal fusion features; The target radio wave propagation model is obtained by adjusting the model parameters of the original radio wave propagation model using a pre-trained reinforcement learning agent model and spatiotemporal fusion features.

[0047] In some embodiments, the method can collect the following parameters every 100ms using onboard sensors, OBU (Onboard Unit), and RSU (Roadside Unit): Vehicle parameters and status: speed v (km / h), acceleration a ( ), yaw angle θ (°); Communication parameters: Received signal strength RSSI (dBm), Doppler frequency shift f d (Hz); Environmental parameters: distance to adjacent vehicles d (m), building height H (m).

[0048] In the above embodiments, the method can optimize the parameters of the original radio wave propagation model based on the spatiotemporal fusion features extracted from real vehicle and road test data, combined with a reinforcement learning agent model, to obtain a target radio wave propagation model that adapts to the actual scenario.

[0049] In some embodiments, feature extraction is performed on the original signal data to obtain spatiotemporal fusion features, including: Building point cloud data and vehicle motion time series data are obtained from the raw signal data; Feature extraction is performed on building point cloud data to obtain target spatial features; The vehicle motion time series data is extracted by a pre-set three-layer bidirectional memory network to obtain the original time series features, and the original time series features are then processed for time synchronization to obtain the target time series features. The spatial features and temporal features of the target are concatenated to obtain the concatenated features. Spatiotemporal fusion features are obtained by adaptively weighting and fusing the spliced ​​features using attention weights.

[0050] In some embodiments, a spatiotemporal feature extraction network is used, i.e., a CNN+LSTM network is used to process time series data.

[0051] In some embodiments, a three-layer convolutional neural network is first used to process building point cloud data. Spatial structure features are extracted using 5×5×5 convolutional kernels, and after batch normalization and ReLU activation, the data enters a max pooling layer. Then, a three-layer bidirectional memory network (LSTM) is applied to process vehicle motion temporal data, with a 12-dimensional input (including velocity, acceleration, and orientation angle) to obtain long-term temporal dependencies. An attention mechanism is then used to fuse spatial and temporal features. First, the point cloud features are flattened into vectors and concatenated with the features from the last time step output by the LSTM. Then, adaptive weighted fusion of features is achieved through attention weight calculation.

[0052] In the above embodiments, the method can extract and fuse spatial and temporal features from the original signal data to generate spatiotemporal fusion features with scene adaptability, thereby improving the input quality of the radio wave propagation model.

[0053] In some embodiments, feature extraction is performed on building point cloud data to obtain target spatial features, including: Feature extraction is performed on building point cloud data to obtain spatial structure features; The spatial structure features are normalized and processed using the ReLU activation function to obtain the first processed features; The first feature is processed by a max pooling layer to obtain the second feature; The second processing feature is evaluated as a vector feature to obtain the target space feature.

[0054] In some embodiments, the method can first use a three-layer convolutional neural network to process building point cloud data, extract spatial structure features through a 5×5×5 convolutional kernel, and then process them through batch normalization and ReLU activation function before entering the max pooling layer.

[0055] In some embodiments, the method can flatten point cloud features into vectors.

[0056] In the above embodiments, the method can extract and process spatial structural features from building point cloud data to obtain high-quality target spatial features suitable for radio wave propagation models.

[0057] In some embodiments, the actions performed by the reinforcement learning agent model include path loss exponent, shadow fading variance, and multipath delay scaling factor. The reinforcement learning agent model also includes a multi-objective reward mechanism, which is constructed based on the absolute error between the predicted value and the actual measured value.

[0058] In some embodiments, the reinforcement learning output model based on the PPO algorithm adjusts the actions. The fused feature vector is used as the input state of the reinforcement learning agent, and the path loss exponent Δn, shadow fading variance Δσ, and multipath delay scaling factor k are used as the actions to be executed. A multi-objective reward mechanism is established, and the reward function r = |predicted RSSI - measured RSSI| is designed based on the absolute error between the predicted value and the actual measured value.

[0059] In the above embodiments, the method can use the path loss exponent, shadow fading variance, and multipath delay scaling factor as the action space, and based on the multi-objective reward mechanism of prediction and measurement errors, to achieve accurate optimization and dynamic adjustment of radio wave propagation model parameters.

[0060] In some embodiments, a lightweight dynamic propagation model is constructed based on the radio wave propagation model, including: A teacher model is constructed in the cloud based on a radio wave propagation model. Knowledge distillation is performed based on the teacher model to obtain a lightweight student model.

[0061] In some embodiments, a 20-layer ResNet teacher model (with more than 10M parameters) is trained in the cloud to output electromagnetic field distribution predictions; a lightweight student model (MobileNetV3 architecture, with less than 100KB parameters) is obtained through knowledge distillation with a temperature coefficient of T=5, and the student model is deployed to the vehicle ECU to achieve real-time prediction.

[0062] In the above embodiments, the method can obtain a lightweight student model adapted to the deployment of the vehicle ECU through knowledge distillation of the cloud-based teacher model, and construct a lightweight dynamic propagation model.

[0063] In some embodiments, generating a target interference source signal based on a pre-built interference source model includes: Based on a pre-defined interference source database and a pre-built interference source model, interference data is synthesized; wherein, the interference source model includes at least a motor broadband noise model and a radar pulse model; Generate the radio frequency signal of the interference source based on the interference data; A channel simulator was used to simulate a multipath fading environment; The target interference source signal is obtained by performing fading processing on the radio frequency signal of the interference source based on the multipath fading environment.

[0064] In some embodiments, this method can utilize MATLAB to model both motor broadband noise and radar pulses. Motor noise can be dynamically generated into a time-varying spectrum based on rotational speed (rpm) and load rate, while radar pulses can have their pulse frequency (PRF) and pulse width configured according to the scene type. Urban roads: 100±20 nanosecond pulse width, 1.0±0.2 kHz repetition frequency, linear frequency modulation; Highway: 50±5 nanosecond pulse width, 5.0±0.5 kHz repetition frequency, phase coding; Tunnel environment: 200±30 nanosecond pulse width, 2.0±0.3 kHz repetition frequency, pulse Doppler modulation.

[0065] In some embodiments, the method may use a Gaussian distribution to generate the center frequency, a Poisson distribution to control the pulse interval, and a parameter selector to dynamically configure the modulation type.

[0066] In some embodiments, the method may employ an NI USRP device to inject two types of interference in the 5.9 GHz band: motor broadband noise and radar pulse, and generate radio frequency signals as follows: sRF(t)=∑ 3 i=1 A i s i (t-τ i )e j2πfit ; Among them, s i This refers to broadband noise from the motor and radar pulses. f i Indicates frequency; A i Represents the coefficient.

[0067] In some embodiments, the method can employ a channel simulator (Keysight S9105A) to simulate a multipath fading environment.

[0068] In some embodiments, the method can establish an on-board OBU radio frequency connection and inject signals into the actual on-board OBU device.

[0069] In some embodiments, the method can run vehicle dynamic scenarios and control the timing of interference injection to achieve real-time control.

[0070] In the above embodiments, the method can comprehensively consider the characteristics of interference sources such as motor broadband noise and radar pulses to generate target interference source signals that closely resemble the real multipath fading environment.

[0071] In some embodiments, electromagnetic compatibility simulation is performed based on a preset traffic scenario simulator and a lightweight dynamic propagation model to obtain simulation results, including: The vehicle dynamics are run through a pre-set traffic scenario simulator. Calculate the scene complexity of vehicle dynamics scenarios; The target interpolation step size and target calculation method are determined based on the preset dynamic resource scheduling strategy and scenario complexity. Simulation calculations were performed based on the target interpolation step size, target calculation method, lightweight student model, and target interference source signal to obtain simulation results.

[0072] In some embodiments, the formula for calculating scene complexity is as follows: C=α•D building +β•v+γ•N vehicle ; Among them, D building Building density; v represents the vehicle speed; N vehicle This represents the number of adjacent vehicles.

[0073] In some embodiments, the method employs a dynamic resource scheduling strategy, which simulates radio wave propagation by changing different interpolation step sizes and calculation methods based on the complexity of the computational scenario, thereby reducing resource waste.

[0074] In some embodiments, if the complexity C < 0.3, i.e., a low-complexity scenario, the interpolation step size is set to 20-50ms, and the ray tracing calculation mode is adopted, which is suitable for open road scenarios. If 0.3≤C<0.6, i.e., medium-complex scene, the interpolation step size is set to 5-20ms, and the MLFMM accelerated calculation mode is adopted, which is suitable for urban street scene; If the complexity C ≥ 0.6, i.e., a highly complex scenario, the interpolation step size is set to 1-5ms, and the full-wave FDTD calculation mode is adopted, which is applicable to tunnel / overpass scenarios.

[0075] In some embodiments, this dynamic conversion mechanism can monitor changes in complexity in real time and automatically save the current state and switch computation modes when a threshold is crossed.

[0076] In the above embodiments, the method can combine traffic scene simulators with dynamic scheduling resources based on scene complexity to determine the appropriate interpolation step size and calculation method, and complete electromagnetic compatibility simulation calculations based on a lightweight student model and target interference source signals.

[0077] In some embodiments, this method constructs a three-dimensional data fusion architecture of "sensing-communication-environment," which can dynamically adjust the propagation model through reinforcement learning, reduce path loss prediction errors and improve model update latency in complex scenarios, thereby reducing simulation resource waste. Simultaneously, by establishing an EMC hardware-in-the-loop platform, it can also achieve dynamic verification of real devices in a controlled electromagnetic environment and propose a dynamic resource adjustment mechanism, improving computational efficiency.

[0078] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions in this application will be clearly and completely described below. In some embodiments, such as Figure 2 As shown, the electromagnetic compatibility simulation method includes: S201. Data Acquisition: Raw signal data is acquired using onboard sensors and transmitted to the RSU edge node.

[0079] S202, Feature Extraction: The feature extractor extracts features from the raw signal data received by the RSU edge nodes to obtain relevant features.

[0080] S203. Model parameter adjustment: The extracted features are input into the reinforcement learning agent, which then adjusts the parameters of the online propagation model based on these features.

[0081] S204 Electromagnetic Compatibility Simulation: The online propagation model with adjusted parameters is input into the dynamic EMC simulation engine, which then performs electromagnetic compatibility simulation.

[0082] S205, Vehicle Control Decision: Make vehicle control decisions based on the simulation results of the dynamic EMC simulation engine.

[0083] In some embodiments, such as Figure 3 As shown, the cloud-vehicle model deployment method includes: S301, Cloud-based training of the teacher model: Training the ResNet-152 teacher model in the cloud.

[0084] S302, Knowledge Distillation Process: S3021. Perform knowledge distillation on the teacher model ResNet-152.

[0085] S3032. In knowledge distillation, on the one hand, temperature scaling (temperature T=5) is used to process the softened target; on the other hand, attention transfer is used to achieve feature alignment, thereby obtaining the student model MobileNetV3.

[0086] S303, Model Quantization: Perform model quantization processing on the student model MobileNetV3.

[0087] S304, Vehicle ECU Deployment and Optimization: S305. Deploy the quantized model onto the vehicle ECU.

[0088] S3051. During the deployment of the vehicle ECU, ECU optimization is performed, including INT8 quantization, operator fusion, and memory optimization.

[0089] In some embodiments, such as Figure 4 As shown, the overall example flow of this method includes: S401, Interference signal generation and injection.

[0090] S4011. Based on the interference source database, interference signals are synthesized using an interference synthesizer.

[0091] S4012. The synthesized interference signal is injected into the OBU (On-Board Unit) under test in radio frequency form via USRP (Universal Software Radio Peripheral).

[0092] S402, Preparation of Dynamic Propagation Model and Electromagnetic Compatibility Simulation.

[0093] S4021, The reinforcement learning agent receives traffic scene information and vehicle sensor data output by the traffic scene simulator.

[0094] S4022, The reinforcement learning agent constructs a dynamic propagation model based on the above inputs.

[0095] S4023, the dynamic propagation model inputs relevant data into the EMC (electromagnetic compatibility) simulation engine to prepare for subsequent simulations.

[0096] S403, Electromagnetic Compatibility Simulation and Integration.

[0097] S4031. Taking into account the situation of the OBU under test with injected interference signals, electromagnetic compatibility simulation is performed on the simulation platform using the EMC simulation engine to obtain simulation results.

[0098] Figure 5 A schematic diagram of an electromagnetic compatibility simulation device is shown. It should be understood that this device is related to... Figure 1The method executed in the middle corresponds to the steps involved in the aforementioned method. The specific functions and effects of the device can be found in the description above. To avoid repetition, detailed descriptions are omitted here.

[0099] The electromagnetic compatibility simulation device includes: The first building unit 510 is used to build a target radio wave propagation model based on reinforcement learning; The second building unit 520 is used to build a lightweight dynamic propagation model based on the radio wave propagation model; Deployment unit 530 is used to deploy the lightweight dynamic propagation model to the on-board ECU of the vehicle to be simulated; The generation unit 540 is used to generate target interference source signals based on a pre-built interference source model; Interference injection unit 550 is used to inject the target interference source signal into the on-board OBU device of the vehicle to be simulated; Simulation unit 560 is used to perform electromagnetic compatibility simulation based on a preset traffic scenario simulator and a lightweight dynamic propagation model to obtain simulation results.

[0100] In some embodiments, the first building unit 510 includes: The first building subunit 511 is used to build the original radio wave propagation model; Acquisition subunit 512 is used to acquire raw signal data through vehicle-mounted sensors, vehicle-mounted communication unit, and road test unit; Extraction subunit 513 is used to extract features from the original signal data to obtain spatiotemporal fusion features; Adjustment subunit 514 is used to adjust the model parameters of the original radio wave propagation model through a pre-trained reinforcement learning agent model and spatiotemporal fusion features to obtain the target radio wave propagation model.

[0101] In some embodiments, the extraction subunit 513 is specifically used to obtain building point cloud data and vehicle motion time series data based on the original signal data; Feature extraction is performed on building point cloud data to obtain target spatial features; The vehicle motion time series data is extracted by a pre-set three-layer bidirectional memory network to obtain the original time series features, and the original time series features are then processed for time synchronization to obtain the target time series features. The spatial features and temporal features of the target are concatenated to obtain the concatenated features. Spatiotemporal fusion features are obtained by adaptively weighting and fusing the spliced ​​features using attention weights.

[0102] In some embodiments, the extraction subunit 513 is specifically used to extract features from building point cloud data to obtain spatial structure features; The spatial structure features are normalized and processed using the ReLU activation function to obtain the first processed features; The first feature is processed by a max pooling layer to obtain the second feature; The second processing feature is evaluated as a vector feature to obtain the target space feature.

[0103] In some embodiments, the actions performed by the reinforcement learning agent model include path loss exponent, shadow fading variance, and multipath delay scaling factor. The reinforcement learning agent model also includes a multi-objective reward mechanism, which is constructed based on the absolute error between the predicted value and the actual measured value.

[0104] In some embodiments, the second building unit 520 includes: The second construction subunit 521 is used to construct a teacher model in the cloud based on the radio wave propagation model; Distillation subunit 522 is used to perform knowledge distillation based on the teacher model to obtain a lightweight student model.

[0105] In some embodiments, the generation unit 540 includes: Synthesis subunit 541 is used to synthesize interference data based on a preset interference source database and a pre-built interference source model; wherein the interference source model includes at least a motor broadband noise model and a radar pulse model. The generation subunit 542 is used to generate an interference source radio frequency signal based on the interference data; Simulation subunit 543 is used to simulate a multipath fading environment using a channel simulator; The fading subunit 544 is used to perform fading processing on the radio frequency signal of the interference source based on the multipath fading environment to obtain the target interference source signal.

[0106] In some embodiments, the simulation unit 560 includes: The running subunit 561 is used to run vehicle dynamic scenarios through a preset traffic scenario simulator; Computational subunit 562 is used to calculate the scene complexity of vehicle dynamic scenes; Subunit 563 is determined to determine the target interpolation step size and target calculation method based on the preset dynamic resource scheduling strategy and scenario complexity. The computational subunit 562 is also used to perform simulation calculations based on the target interpolation step size, target calculation method, lightweight student model, and target interference source signal to obtain simulation results.

[0107] like Figure 6As shown, this application provides an electronic device 600, which includes a processor 601 and a memory 602. The processor 601 and the memory 602 are interconnected and communicate with each other through a communication bus 603 and / or other forms of connection mechanism (not shown). The memory 602 stores a computer program that can be executed by the processor 601. When the computing device is running, the processor 601 executes the computer program to perform the method in any of the aforementioned optional implementations.

[0108] This application provides a computer-readable storage medium storing a computer program, which, when executed by a processor, performs the method in any of the aforementioned optional implementations.

[0109] The computer-readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0110] This application provides a computer program product, which includes a computer program that, when run by a processor, executes the method in any of the aforementioned optional implementations.

[0111] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and not to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. These modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application, and they should all be covered within the scope of the claims and specification of this application. In particular, as long as there is no conflict, the various technical features mentioned in the embodiments can be combined in any way. This application is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. An electromagnetic compatibility simulation method, characterized in that, include: Construct a target radio wave propagation model based on reinforcement learning; A lightweight dynamic propagation model is constructed based on the aforementioned radio wave propagation model; The lightweight dynamic propagation model is deployed on the vehicle ECU of the vehicle to be simulated; Generate target interference source signals based on a pre-built interference source model; The target interference source signal is injected into the on-board unit (OBU) of the vehicle to be simulated, and electromagnetic compatibility simulation is performed according to the preset traffic scenario simulator and the lightweight dynamic propagation model to obtain the simulation results.

2. The electromagnetic compatibility simulation method according to claim 1, characterized in that, The construction of the target radio wave propagation model based on reinforcement learning includes: Construct the original radio wave propagation model; Raw signal data is acquired through onboard sensors, onboard communication units, and road test units; Feature extraction is performed on the original signal data to obtain spatiotemporal fusion features; The target radio wave propagation model is obtained by adjusting the model parameters of the original radio wave propagation model using a pre-trained reinforcement learning agent model and the spatiotemporal fusion features.

3. The electromagnetic compatibility simulation method according to claim 2, characterized in that, The step of extracting features from the original signal data to obtain spatiotemporal fusion features includes: Based on the original signal data, obtain building point cloud data and vehicle motion time sequence data; Feature extraction is performed on the building point cloud data to obtain the target spatial features; The vehicle motion time series data is feature extracted by a preset three-layer bidirectional memory network to obtain the original time series features, and the original time series features are time synchronized to obtain the target time series features. The target spatial features and the target temporal features are concatenated to obtain concatenated features; The spliced ​​features are adaptively weighted and fused using attention weights to obtain spatiotemporal fusion features.

4. The electromagnetic compatibility simulation method according to claim 3, characterized in that, The step of extracting features from the building point cloud data to obtain target spatial features includes: Feature extraction is performed on the building point cloud data to obtain spatial structure features; The spatial structure features are normalized and processed using the ReLU activation function to obtain the first processed features; The first processed feature is processed by a max pooling layer to obtain the second processed feature; The second processed feature is evaluated as a vector feature to obtain the target space feature.

5. The electromagnetic compatibility simulation method according to claim 2, characterized in that, The actions performed by the reinforcement learning agent model include path loss exponent, shadow fading variance, and multipath delay scaling factor. The reinforcement learning agent model also includes a multi-objective reward mechanism, which is constructed based on the absolute error between the predicted value and the actual measured value.

6. The electromagnetic compatibility simulation method according to claim 1, characterized in that, The construction of a lightweight dynamic propagation model based on the aforementioned radio wave propagation model includes: A teacher model is constructed in the cloud based on the described radio wave propagation model; Knowledge distillation is performed based on the teacher model to obtain a lightweight student model.

7. The electromagnetic compatibility simulation method according to claim 1, characterized in that, The step of generating the target interference source signal based on the pre-built interference source model includes: Interference data is synthesized based on a pre-defined interference source database and a pre-built interference source model; wherein the interference source model includes at least a motor broadband noise model and a radar pulse model. An interference source radio frequency signal is generated based on the interference data; A channel simulator was used to simulate a multipath fading environment; The interference source radio frequency signal is subjected to fading processing based on the multipath fading environment to obtain the target interference source signal.

8. The electromagnetic compatibility simulation method according to claim 6, characterized in that, The electromagnetic compatibility simulation is performed based on a preset traffic scenario simulator and the lightweight dynamic propagation model to obtain simulation results, including: The vehicle dynamics are run through a pre-set traffic scenario simulator. Calculate the scene complexity of the vehicle dynamic scene; The target interpolation step size and target calculation method are determined based on the preset dynamic resource scheduling strategy and the complexity of the scenario. Simulation results are obtained by performing simulation calculations based on the target interpolation step size, the target calculation method, the lightweight student model, and the target interference source signal.

9. An electromagnetic compatibility simulation device, characterized in that, The electromagnetic compatibility simulation device includes: The first building unit is used to construct a target radio wave propagation model based on reinforcement learning; The second building unit is used to build a lightweight dynamic propagation model based on the radio wave propagation model. The deployment unit is used to deploy the lightweight dynamic propagation model to the on-board ECU of the vehicle to be simulated; The generation unit is used to generate target interference source signals based on a pre-built interference source model; An interference injection unit is used to inject the target interference source signal into the on-board OBU device of the vehicle to be simulated. The simulation unit is used to perform electromagnetic compatibility simulation based on a preset traffic scenario simulator and the lightweight dynamic propagation model, and obtain simulation results.

10. An electronic device, characterized in that, The electronic device includes a memory and a processor, the memory being used to store a computer program, and the processor running the computer program to cause the electronic device to perform the electromagnetic compatibility simulation method according to any one of claims 1 to 8.

11. A readable storage medium, characterized in that, The readable storage medium stores a computer program, which, when executed by a processor, performs the electromagnetic compatibility simulation method according to any one of claims 1 to 8.

12. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, performs the electromagnetic compatibility simulation method according to any one of claims 1 to 8.

Images