Electric vehicle electromagnetic radiation monitoring method and system

Abstract

The application belongs to the field of electric vehicle detection, and provides an electromagnetic radiation monitoring method and system for electric vehicles, which comprises arranging electromagnetic radiation sensing nodes at multiple preset positions inside and outside the electric vehicle; an electromagnetic cognition modeling system with multiple source inputs is established to generate a graph model for describing the distribution evolution law of electromagnetic radiation; a coupling model between multiple electromagnetic source modules inside the electric vehicle is constructed in the electromagnetic cognition modeling system, and an induced risk score is introduced into the graph model as a correction factor; based on the graph model, a high radiation area that may occur in the electric vehicle under the current working condition is predicted, and the dynamic sampling priority of each sensing node is determined; according to the sampling priority of each sensing node and the local real-time monitoring result, the sampling frequency is autonomously adjusted, and the local enhanced sampling of the predicted high-risk area is realized through the inter-node cooperation mechanism.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of electric vehicle detection, and in particular relates to an electric vehicle electromagnetic radiation monitoring method and system. BACKGROUND

[0002] With the rapid development of the electric vehicle industry, the degree of electrification of the whole vehicle has significantly improved. Various electronic modules such as drive motors, high-frequency inverters, on-board chargers, BMS systems, and on-board communication terminals work continuously during vehicle operation, making the electromagnetic radiation environment of the whole vehicle increasingly complex. Electric vehicles may generate electromagnetic waves of different frequencies and intensities under various operating conditions such as acceleration, braking, shifting, charging, and high-speed driving. These electromagnetic waves not only may interfere with the vehicle's electronic systems, but also may affect the electromagnetic exposure safety of the passengers.

[0003] Most existing electromagnetic radiation monitoring schemes are based on fixed point distribution and equal periodic sampling methods, which cannot dynamically adapt to the changing operating conditions of electric vehicles. Moreover, they usually ignore the electromagnetic coupling relationship between various electromagnetic source modules in the vehicle, and cannot effectively identify the nonlinear radiation enhancement phenomenon caused by module coordination, frequency resonance, or structural coupling. In addition, although high-frequency continuous sampling improves the timeliness of monitoring, it also brings problems such as high power consumption, strong interference, and data redundancy, which are not conducive to the long-term stable operation of the system. Therefore, there is an urgent need for an electric vehicle electromagnetic radiation dynamic monitoring method that can integrate multi-source input data, implement spatiotemporal perception intelligent modeling, and have a local enhancement response capability, to improve the identification accuracy of high-risk areas and abnormal behaviors, while reducing unnecessary monitoring costs and improving the electromagnetic safety management level of the whole vehicle. SUMMARY

[0004] To solve the problems in the prior art, the present application provides an electric vehicle electromagnetic radiation monitoring method, comprising the following steps:

[0005] Electromagnetic radiation sensing nodes are arranged at multiple predetermined positions inside and outside the electric vehicle. Each sensing node is used to collect electromagnetic radiation time domain signals and frequency domain features at the corresponding position.

[0006] An electromagnetic cognitive modeling system with multiple input sources is established. The electromagnetic cognitive modeling system integrates multi-dimensional input variables such as vehicle operating parameters, vehicle electronic module states, passenger distribution information, and external environmental interference background, and generates a graph model for describing the evolution law of electromagnetic radiation distribution.

[0007] A coupling model between multiple electromagnetic source modules inside the electric vehicle is constructed in the electromagnetic cognitive modeling system. An electromagnetic induction causal relationship graph is established to identify potential nonlinear radiation enhancement events caused by module coordination or resonance effects, and the induced risk score is introduced as a correction factor into the graph model.

[0008] Based on the graph model, the high radiation areas that may occur in the electric vehicle under the current operating conditions are predicted, and the dynamic sampling priority of each sensor node is determined. The priority is determined by the electromagnetic change rate, spatial radiation gradient and inter-module coupling-induced risk level of the node area.

[0009] According to the sampling priority of each sensor node and the local real-time monitoring results, the sampling frequency is adjusted autonomously, and local enhanced sampling of predicted high-risk areas is achieved through the inter-node collaboration mechanism.

[0010] Furthermore, sensing nodes are arranged in the battery pack area, the drive motor area, the periphery of the power electronic module, the vehicle communication system location, the charging interface area, the key locations of the passenger compartment and the external reference points of the vehicle body.

[0011] Furthermore, the nodes in the graph model include state nodes, space nodes, module nodes and risk nodes. The model is trained using a graph neural network and outputs electromagnetic radiation intensity prediction values ​​and risk scores based on the temporal evolution information and structural connection relationships of each node.

[0012] Furthermore, the inter-module coupling model includes: electrical direct coupling path, spatial field coupling path and frequency modulation coupling path. The coupling path is represented in the form of a graph structure, where the edge weight is used to characterize the coupling strength or resonance probability between modules and participates in the construction of the electromagnetic induced causal association graph.

[0013] Furthermore, each sensor node includes a local trend detection module and a collaborative perception module. The local trend detection module is used to determine whether to up-sample based on the electromagnetic signal fluctuation within a short time window, and the collaborative perception module is used to broadcast the abnormal status to neighboring nodes to achieve multi-node joint response sampling in high-risk areas.

[0014] Another aspect of the present invention provides an electric vehicle electromagnetic radiation monitoring system, comprising the following modules:

[0015] Multiple electromagnetic radiation sensing nodes are arranged at multiple preset locations inside and outside the electric vehicle, and each sensing node is used to collect the time domain signal and frequency domain characteristics of electromagnetic radiation at the corresponding location;

[0016] The electromagnetic cognitive modeling module is used to integrate multi-dimensional input variables including vehicle operating parameters, on-board electronic module status, occupant distribution information, and external environmental interference background to establish a graphical model describing the evolution of electromagnetic radiation distribution;

[0017] A coupling analysis module, provided in the electromagnetic cognitive modeling system, is used to construct a coupling model between multiple electromagnetic source modules within the electric vehicle and establish an electromagnetic induced causal association diagram to identify potential nonlinear radiation enhancement events caused by module synergy or resonance effects, and introduce an induced risk score as a correction factor into the diagram model;

[0018] A priority assessment module is used to predict the high radiation areas that may appear in the electric vehicle under the current operating conditions based on the graph model, and determine the dynamic sampling priority of each sensor node. The priority is determined by the electromagnetic change rate of the node area, the spatial radiation gradient, and the risk level of coupling induced between modules;

[0019] The frequency adjustment and collaborative sampling module is used to autonomously adjust the sampling frequency according to the sampling priority of each sensor node and the local real-time monitoring results, and to achieve local enhanced sampling of predicted high-risk areas through an inter-node collaboration mechanism.

[0020] Furthermore, electromagnetic radiation sensing nodes are set in the battery pack area, the drive motor area, the periphery of the power electronic module, the vehicle communication system location, the charging interface area, the key locations of the passenger compartment and the external reference points of the vehicle body.

[0021] Furthermore, the nodes in the graph model include state nodes, space nodes, module nodes and risk nodes. The graph model is trained through a graph neural network and outputs electromagnetic radiation intensity prediction values ​​and risk scores based on the temporal evolution information and structural connection relationship of each node.

[0022] Furthermore, the inter-module coupling model includes: electrical direct coupling path, spatial field coupling path and frequency modulation coupling path. The coupling path is represented in the form of a graph structure, where the edge weight is used to characterize the coupling strength or resonance probability between modules and participates in the construction of the electromagnetic induced causal association graph.

[0023] Furthermore, each sensor node includes a local trend detection module and a collaborative perception module. The local trend detection module is used to determine whether to up-sample based on the electromagnetic signal fluctuation within a short time window, and the collaborative perception module is used to broadcast the abnormal status to neighboring nodes to achieve multi-node joint response sampling in high-risk areas.

[0024] This invention proposes a method and system for monitoring electromagnetic radiation in electric vehicles. By deploying electromagnetic radiation sensing nodes at multiple key locations inside and outside the vehicle, an electromagnetic cognitive modeling system is constructed based on multi-source state inputs. Furthermore, a graph model and module coupling mechanism are introduced. This method effectively overcomes the shortcomings of existing technologies in monitoring response, modeling capabilities, and energy consumption control. The system has the following beneficial technical effects:

[0025] By adopting graph neural networks and heterogeneous graph structures, integrating multi-source inputs such as vehicle operating status, module working information, occupant distribution and external interference, a spatiotemporal joint modeling system is established, which can accurately predict the spatial distribution and intensity change trend of electromagnetic radiation under different working conditions.

[0026] By constructing an electromagnetic coupling model and induced causal diagram between modules, we can actively identify the nonlinear radiation enhancement behavior caused by module collaboration or frequency resonance, predict and intervene in potential high-risk events in advance, and improve the system's electromagnetic compatibility protection capabilities.

[0027] The sampling priority of each sensor node is dynamically generated based on the graph reasoning results and risk scores. Combined with local trend detection and inter-node coordination mechanism, local enhanced sampling in high-risk areas and intermittent or dormant control in low-risk areas are achieved, taking into account both monitoring accuracy and power consumption optimization.

[0028] Through continuous feedback and online update mechanisms, the system can continuously optimize the map structure and parameter weights during long-term operation, adapt to the monitoring needs of different vehicle models, different module layout structures and different operating environments, and has strong versatility and scalability. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0030] Figure 1 is a flow chart of the method of the present invention;

[0031] Figure 2 It is a system block diagram of the present invention. DETAILED DESCRIPTION

[0032] Below, the invention is preferably described with reference to the accompanying drawings and specific embodiments.

[0033] This embodiment solves the above problem through the following steps:

[0034] In one embodiment, reference Figure 1The present invention provides an electromagnetic radiation monitoring method for electric vehicles, which performs real-time monitoring, intelligent scheduling and dynamic optimization of electromagnetic radiation generated under different operating conditions. The method is suitable for continuous perception and spatiotemporal feature analysis of the electromagnetic radiation intensity that may appear in the key components inside the electric vehicle (including battery packs, drive motors, power electronic modules, charging systems and on-board communication devices) and the passenger environment inside and outside the vehicle. Through the collaborative work of multiple sensor nodes and intelligent control strategies, it can achieve key monitoring of high-risk areas and power consumption control in low-change areas, thereby ensuring the electromagnetic radiation safety and system stability of electric vehicles during market testing and actual operation.

[0035] The method specifically comprises the following steps:

[0036] Step S1 : deploying electromagnetic radiation sensing nodes at multiple preset locations inside and outside the electric vehicle, wherein each sensing node is used to collect the time domain signal and frequency domain characteristics of electromagnetic radiation at the corresponding location.

[0037] Specifically, electromagnetic radiation sensing nodes are deployed at multiple preset key locations inside and outside the electric vehicle to perform distributed, differentiated, dynamic perception of the electromagnetic radiation signals generated by the vehicle under various operating conditions; each sensing node collects the electromagnetic radiation time domain signal (such as the curve of the change of electric field intensity over time) and frequency domain characteristics (such as spectrum energy distribution, main frequency power density, harmonic amplitude, etc.) at its location, and uploads the collected data to the central processing module for subsequent modeling and scheduling.

[0038] The preset locations include but are not limited to the following areas:

[0039] Battery pack area: Sensors are deployed to monitor low-frequency radiation generated during high-current charging and discharging;

[0040] Near the drive motor: used to capture medium-frequency electromagnetic signals caused by motor commutation and rotating magnetic field;

[0041] Around power electronic modules (such as IGBT modules): monitoring high-frequency harmonic interference and transient electromagnetic pulses;

[0042] Near vehicle communication systems (such as Wi-Fi, Bluetooth, and V2X antennas): monitor high-frequency continuous waves generated during communication;

[0043] Charging interface area: Evaluate pulse electromagnetic interference during high voltage connection and disconnection;

[0044] Key locations in the passenger compartment, such as the main driver's seat and child seat area: used to monitor radiation intensity and dose exposure levels in areas close to the human body;

[0045] External reference point of the vehicle body: Set the reference sensor for background noise calibration and external interference identification.

[0046] Preferably, each sensor node can adopt a broadband vector electromagnetic sensor module with dual-channel perception capabilities of electric and magnetic fields, support multi-band sensing in the range of 10kHz to 1GHz, and have local preliminary processing and data compression capabilities to adapt to subsequent graph model construction and sampling priority scheduling.

[0047] Step S2: Establish an electromagnetic cognitive modeling system with multi-source input. The electromagnetic cognitive modeling system integrates multi-dimensional input variables including vehicle operating parameters, vehicle electronic module status, occupant distribution information, and external environmental interference background to generate a graph model for describing the evolution law of electromagnetic radiation distribution.

[0048] As highly electrified mobile systems, electric vehicles (EVs) generate electromagnetic radiation not from a single source but from the combined operation of multiple electrical modules (such as the drive motor, IGBT module, charging system, and onboard communication unit) under varying operating conditions. These radiation behaviors exhibit significant dynamic and nonlinear characteristics across time, frequency, and space. In particular, during EV operation, factors such as the coupling between different modules, switching between operating states, and changes in modulation methods complicate radiation behavior, making it difficult for traditional monitoring methods based on static rules or empirical models to accurately predict high-risk areas.

[0049] Therefore, to achieve intelligent monitoring and proactive identification of electromagnetic radiation from electric vehicles, it is necessary to establish an electromagnetic cognitive modeling system that integrates multiple input data and possesses self-learning and reasoning capabilities. By introducing a graph-structured modeling approach, complex information such as vehicle operating conditions, module status, and spatial structure can be organized into a learnable graph. This allows for dynamic modeling, inference prediction, and risk scoring of radiation distribution, providing decision support for subsequent sampling strategy optimization.

[0050] The specific steps are as follows:

[0051] Step 21: Multi-source data collection and structured modeling

[0052] First, the system extracts input variables of multiple dimensions from the vehicle network, including but not limited to:

[0053] Vehicle operating parameters: such as current speed, acceleration, motor current, battery voltage, and motor temperature;

[0054] Module working status: such as IGBT duty cycle, PWM frequency, BMS charge and discharge status;

[0055] Occupant distribution information: Determine seat usage status through pressure sensors and infrared / visual systems;

[0056] External interference information: Collects information such as the electromagnetic background power spectrum density and wireless communication interference level outside the vehicle.

[0057] The above data are synchronized in the time dimension and normalized in a standardized form to construct the input tensor and state description vector for subsequent modeling.

[0058] Step 22: Construct the topology of the electromagnetic spectrum model

[0059] Based on the normalized data, an electromagnetic spectrum model is constructed, including:

[0060] Define node types: such as "battery status node", "current peak node", "occupant position node", "radiation risk node", etc.;

[0061] Define the semantic relationship of edges: for example, "motor speed ↑ → PWM frequency ↑ → intermediate frequency radiation ↑", and model it as a weighted directed edge;

[0062] Add spatial edges (such as sensor point adjacency), module coupling edges (such as the feedback path between IGBT and BMS), and time edges (for modeling state evolution) to the graph;

[0063] Construct a heterogeneous graph structure, initialize the graph neural network on it, and set the initial node embedding vector.

[0064] Step 23: Training and iterative learning

[0065] Graph neural networks (such as GCN, GAT, or GraphSAGE) are used for training. The input is the constructed graph structure and the corresponding historical radiation data labels. The output is:

[0066] The predicted value of radiation intensity of each node;

[0067] Radiation risk scores for each area within the vehicle;

[0068] Spatial tracing path of potential abnormal radiation events.

[0069] During the training process, supervised learning strategies (such as minimizing the mean square error between predicted and measured values) or semi-supervised strategies (combined with graph structure propagation mechanisms when labels are missing) can be used.

[0070] Step 24: Real-time reasoning and scheduling decision support

[0071] The trained model is deployed in the vehicle's central processing module. During actual operation, new vehicle status data is input in real time, and the following results are output through the forward reasoning process:

[0072] Prediction diagram of electromagnetic radiation spatial distribution under current working conditions;

[0073] High-risk area identification results and credibility scores;

[0074] The sampling priority of each sampling node is recommended to guide subsequent sampling frequency adjustment and power consumption management.

[0075] This step overcomes the limitations of traditional monitoring methods, which rely solely on local sensor perception and static rule-based reasoning, by introducing multi-source state perception and graph modeling. First, the system can perceive the potential coupling relationships and inducing mechanisms between modules, enabling high-risk identification at the module behavior level. Second, by introducing structured graph representation and graph neural network reasoning mechanisms, it can identify the generation paths and spatial propagation trends of electromagnetic anomalies in high-dimensional nonlinear state space, demonstrating excellent generalization capabilities. Third, the model possesses adaptive learning capabilities, continuously optimizing its prediction performance as vehicle data accumulates, making it suitable for self-learning evolution across different vehicle models and operating scenarios.

[0076] Step S3, constructing a coupling model between multiple electromagnetic source modules inside the electric vehicle in the electromagnetic cognitive modeling system, establishing an electromagnetic induced causal association diagram for identifying potential nonlinear radiation enhancement events caused by module synergy or resonance effects, and introducing the induced risk score as a correction factor into the spectrum model.

[0077] In step S2, an electromagnetic spectrum model was constructed that integrates multi-source data, including vehicle operating status, module operating conditions, and external interference, to describe the spatiotemporal distribution of electromagnetic radiation under different operating conditions of electric vehicles. However, in actual electric vehicle systems, some sudden electromagnetic radiation anomalies are not driven by a single module independently, but often arise from the nonlinear amplification effects induced by the coordinated operation or frequency coupling of multiple electromagnetic source modules.

[0078] For example, if control frequency synchronization occurs between the electric drive system and power devices, it may trigger phenomena such as harmonic overlap and resonant excitation, resulting in a sudden increase in radiated power in certain areas. Because this type of behavior is hidden, sudden, and highly nonlinear, it is difficult to accurately capture it by relying solely on node state reasoning in traditional graphs. Therefore, it is necessary to further introduce inter-module electromagnetic coupling models and induced causal graph structures based on graph modeling to identify these complex inducing mechanisms. The corresponding risk scoring results can then be introduced into the graph reasoning process to improve the ability to predict abnormal events.

[0079] The specific implementation process includes:

[0080] Step S31: Extract the main electromagnetic source modules and their electromagnetic interaction characteristics

[0081] Based on the system's electrical schematics and module control logic, we extracted core modules with independent electromagnetic emission capabilities, such as the motor controller (MCU), IGBT power switch unit, BMS system, battery pack, radar, and V2X communication unit. We also developed a characteristic description for each module, including its typical electromagnetic emission frequency band, modulation strategy, and harmonic behavior.

[0082] Step S32: Constructing an electromagnetic coupling path model between modules

[0083] Based on the actual line connection, electrical coupling conditions and spatial arrangement, an electromagnetic coupling relationship network between modules is established, including:

[0084] Electrical direct coupling paths: common power bus, ground return, control signal link;

[0085] Spatial field coupling paths: such as close wiring, metal cavity reflection coupling, and capacitive induction interference;

[0086] Frequency / modulation coupling path: such as PWM frequency synchronization, modulation side band overlap, etc.

[0087] The coupling network is represented by a graph structure, where nodes are modules, edges are coupling relationships, and edge weights represent coupling strength or historical resonance probability.

[0088] Step S33: Create an induced causal diagram and define high-risk combination patterns

[0089] Based on the above coupling diagram, through historical operation data or simulation data analysis:

[0090] Identify the statistical correlation between the state of a specific module combination (such as A frequency ↑ & B duty cycle ↑) and electromagnetic anomaly events (E field intensity surge) at the target sensor node;

[0091] Use causal reasoning methods (such as Bayesian networks and frequent item set mining) to construct induced causal graphs;

[0092] A logical mapping is established between each module combination state and potential anomalies, marked as a high-risk combination mode, and the corresponding induction probability value is set.

[0093] Step S34: Generate induced risk score in real time

[0094] During vehicle operation, the system detects the working status of each module in real time and matches it with the high-risk combination patterns identified in the induced causal diagram:

[0095] If the current module state is highly similar to a known combination, a risk score between 0 and 1 is output;

[0096] This score reflects the possibility that the current system is on the verge of induced radiation enhancement;

[0097] Simultaneously record the region where interference amplification is likely to occur on the coupling path.

[0098] Step S35: Take the risk score as a graph correction factor

[0099] Introduce the induced risk score into the graph model constructed in step S2 as a correction factor:

[0100] If there is a high risk score in a certain region, increase the radiation intensity prediction value in the graph of that region;

[0101] In the sampling scheduling module, increase the priority of the corresponding node;

[0102] If there is a high-risk propagation path across nodes, the adjacent region nodes can also be activated in advance to prevent in advance.

[0103] This step enables the original graph model to have self-awareness of induced risks within the system, breaking the limitation of traditional graph modeling relying only on static state input. After introducing the coupling model and causal reasoning structure, the system can not only judge the current risk based on module state, but also predict the upcoming anomaly based on the coupling trend between modules, having a certain predictive foresight. In addition, the induced risk score as a dynamic correction in graph reasoning significantly enhances the electromagnetic anomaly response capability of the model under the influence of multiple modules, improving the adaptability of the system to complex scenarios and the monitoring sensitivity at high-risk moments.

[0104] Exemplarily:

[0105] In a certain high-speed running test, the vehicle enters the slope acceleration stage, and the following module states are detected:

[0106] The IGBT module switching frequency rises to 26 kHz, and the duty cycle reaches 90%;

[0107] The electric drive motor speed is increased to 6000 rpm;

[0108] The radar system is in active detection state, and the V2X module periodically sends broadcast frames;

[0109] The instantaneous discharge current of the battery exceeds 220A.

[0110] The system judges that the current overall radiation level is in the medium risk interval in the S2 stage graph model, but through the module coupling model and induced causal graph constructed in S3, it further identifies that the above state combination is a high-risk inducing combination of historical medium-frequency electromagnetic anomaly events, and calculates the induced risk score as 0.94.

[0111] Based on this, the atlas model corrected the radiation prediction value of the corresponding area, marked the front of the passenger compartment and the top of the vehicle's motor compartment as high-risk points, and the system dispatched the sensors in this area to high-frequency sampling mode.

[0112] Step S4, based on the graph model, predict the high radiation areas that may appear in the electric vehicle under the current operating conditions, and determine the dynamic sampling priority of each sensor node. The priority is jointly determined by the electromagnetic change rate, spatial radiation gradient and inter-module coupling-induced risk level of the area where the node is located.

[0113] In step S2, the electromagnetic cognitive modeling system integrates data such as vehicle operating status, module operating information, occupant location, and external environment to construct a spectral model to characterize the overall electromagnetic radiation distribution trend of the electric vehicle. In step S3, the electromagnetic coupling mechanism between modules is introduced to identify and quantify possible nonlinear enhanced radiation events, and output a risk score. The spectral model and coupled risk score together constitute a systematic understanding of the electromagnetic radiation distribution and abnormal probability under the current operating conditions.

[0114] However, sensor node resources are limited, and high-frequency sampling increases power consumption and background interference. To balance monitoring accuracy and system stability, a sampling priority scheduling mechanism is needed. This mechanism dynamically adjusts the operating state of each node in a risk-driven manner, aligning its sampling behavior with the actual risk situation. Specifically, the predictions output by the aforementioned model should be converted into sampling control variables, prioritizing enhanced sensing coverage in high-risk areas and reducing ineffective sampling in low-variability areas, ultimately achieving optimal intelligent sensing of the vehicle's electromagnetic state.

[0115] Specifically:

[0116] Step S41: Input multi-source model output data

[0117] The system takes the following information output by the graph model at the previous moment as input:

[0118] The predicted value of electromagnetic radiation in each spatial area of ​​the entire vehicle at the current moment;

[0119] The electromagnetic change rate of each sensor node's location (i.e., the difference between the previous and next predicted values);

[0120] Spatial gradient, i.e., the spatial first-order derivative of the radiation intensity in adjacent areas;

[0121] The module-induced risk score values ​​are calculated in S3.

[0122] Step S42: Constructing a sampling priority function

[0123] For each sensor node i, construct its sampling priority function P i, this function is a multi-factor weight model:

[0124]

[0125] in, Indicates the electromagnetic change rate in the area where the node is located;

[0126] Represents the spatial radiation gradient (the difference in electromagnetic intensity with adjacent nodes);

[0127] R i represents the induced risk score obtained in step S3;

[0128] α, β, and γ represent adjustable weighting coefficients used to adapt to different task requirements (such as safety monitoring priority, power optimization priority, etc.).

[0129] Step S43: Priority normalization and level division

[0130] To facilitate scheduling and control, the system sets the P i

[0131] The values ​​are normalized and divided into multiple priority levels (such as 5 levels):

[0132] High priority (activation + high-frequency sampling);

[0133] Medium-high priority (activation + intermediate frequency sampling);

[0134] Medium level (maintain current frequency);

[0135] Medium to low level (entering intermittent sampling mode);

[0136] Low priority (enter sleep / shutdown state).

[0137] Furthermore, the classification can be achieved by setting a threshold or based on a clustering algorithm (such as K-means).

[0138] Step S44: Generate sampling scheduling instructions

[0139] Each node is assigned the following behavior parameters based on its priority level:

[0140] Sampling frequency (e.g., 100 Hz, 10 Hz, 1 Hz, 0.1 Hz);

[0141] Sampling window length (affects data granularity and compression ratio);

[0142] Activation cycle (such as periodic wakeup or event-driven wakeup);

[0143] Whether to perform relay transmission (affects data upload frequency and energy consumption);

[0144] The scheduling command is sent to each sensor node through the bus and executed by its local control module.

[0145] This step transforms the electromagnetic spectrum model and module coupling risk score results into a sampling control strategy, achieving a direct linkage from data recognition to hardware scheduling. On the one hand, this increases monitoring density in high-risk areas, enabling rapid detection and response to sudden radiation anomalies. On the other hand, by controlling the dormancy of low-risk areas, it significantly reduces overall system power consumption and redundant data processing, extending the system's operational life and improving overall monitoring efficiency.

[0146] In addition, the mechanism has good adaptability and can dynamically adjust the sampling strategy according to changes in the vehicle's operating status. It does not rely on fixed rules, but is based on real-time regulation based on risk perception, making it suitable for flexible deployment on different platforms and under different operating conditions.

[0147] For example:

[0148] During a test on urban road conditions, the vehicle entered an accelerated merging state. The atlas model predicted that the radiation level near the front cabin of the vehicle and in the area directly in front of the driver in the vehicle would rise rapidly, with the rate of change being 3.6 times the average of the previous 10 seconds, and the spatial gradient would be significantly enhanced. At the same time, the system determined that the IGBT and motor controller frequencies were synchronized, induced a risk score as high as 0.88.

[0149] Based on the above data, the priority score of the sensor node in the middle of the front of the vehicle is calculated to be 0.91. The system schedules it to the high priority level, assigns a sampling frequency of 100Hz, and continuously records the EM interference waveform. The node located under the co-pilot's feet is only assigned a low-frequency sampling of 0.5Hz due to its slow changes, gentle gradient, and no risk score, and is set to enter intermittent sleep mode.

[0150] Subsequent measurements showed that the high-frequency sampling node successfully captured a 1.5-second electromagnetic anomaly pulse event with typical medium-frequency harmonic superposition characteristics, confirming the effectiveness of the scheduling mechanism; while the low-priority node still maintained the continuity of basic background data in an energy-saving state.

[0151] In step S5, the sampling frequency is adjusted autonomously according to the sampling priority of each sensor node and the local real-time monitoring results, and local enhanced sampling of the predicted high-risk area is achieved through the inter-node coordination mechanism.

[0152] In step S4, the system calculates the sampling priority for each sensor node based on the graph model and the induced risk score, completing the global scheduling plan. However, the actual monitoring environment is dynamic, and some areas may enter a high-risk state in a very short period of time due to sudden interference, abnormal status, module failure, etc. Therefore, relying solely on global priority settings may miss critical anomalies during the data refresh interval.

[0153] To enhance the system's dynamic response capabilities, each sensor node must possess intelligent local perception and autonomous adjustment capabilities. This means that after receiving an initial priority, it can dynamically fine-tune its sampling frequency based on the short-term trends of its own sampled data. Furthermore, because electromagnetic interference propagates spatially, a collaborative mechanism should be established between adjacent nodes. When a localized radiation increase is detected, the sampling frequencies of multiple nodes should be jointly increased, enabling enhanced sampling in high-risk areas to ensure coverage density and data integrity.

[0154] Specifically, the implementation process includes the following steps:

[0155] Step S51: Initialize sampling frequency

[0156] Each sensor node receives the priority parameters and recommended sampling frequency f0 issued by the central system in step S4 and enters the sampling initialization phase at this frequency. The sampling module begins to regularly collect and cache local electromagnetic waveform data (electric field strength, electromagnetic spectrum, etc.).

[0157] Step S52: Local trend detection and adaptive sampling frequency adjustment

[0158] Each node is equipped with a lightweight trend analysis algorithm (such as sliding average, variance detection or short-time Fourier transform) to locally determine the change of electromagnetic signal characteristics. If one of the following situations occurs in N consecutive samplings:

[0159] The fluctuation amplitude of electric field intensity increases significantly (for example, exceeds 2 standard deviations of the mean);

[0160] The energy in the high frequency band suddenly increases or new frequency components appear;

[0161] The duration of the abnormal signal exceeds the preset threshold;

[0162] The node triggers the "sampling frequency adaptive adjustment mechanism" to increase the current frequency f to f'=f0×(1+δ), where

[0163] δ is the amplification factor (e.g., 0.5 to 2.0), which enables enhanced local responsiveness.

[0164] Step S53: Inter-node collaborative perception mechanism

[0165] Each node regularly broadcasts its current monitoring status (such as average electric field strength, frequency characteristic change value, and whether anomalies are detected) to neighboring nodes, forming a local state propagation subnetwork.

[0166] When a node A detects a rising trend in local risk, it sends a collaborative activation signal to its neighboring nodes B and C, prompting them to enter a collaborative warning state. Based on this, the neighboring nodes can determine whether they need to:

[0167] Upsampling in advance;

[0168] Activate standby low-power nodes;

[0169] Extend the sampling window time to obtain more stable spectral characteristics.

[0170] This mechanism realizes a rapid closed loop of local radiation events → multi-node joint response.

[0171] Step S54: Dynamic feedback and global strategy update

[0172] After each sampling cycle, each node uploads its frequency adjustment history, local exception capture records, and collaborative feedback information to the central system for correction of graph model input or recalculation of priority, thereby entering the scheduling closed loop for the next cycle.

[0173] This system has achieved a transition from "global static scheduling" to "local dynamic response," building a real-time sampling and control system based on risk perception. It can rapidly increase sampling density during periods of abnormalities and sudden incidents to capture subtle signal characteristics. It only increases sampling frequency in necessary areas, while maintaining low power consumption in other areas to avoid overall performance loss. Thanks to a node coordination mechanism, even if individual nodes misdetect or go dormant, they can leverage surrounding nodes for additional sensing, improving the system's fault tolerance. This mechanism is suitable for densely deployed sensor systems and can be extended to sparsely deployed structures through soft coordination.

[0174] For example:

[0175] During a complete vehicle immunity test on an electric vehicle, the vehicle entered charging mode and simultaneously activated its radar module. At this point, the sensor node deployed near the charging port operated at a 10Hz frequency. Within a short period of time, two strong intermediate-frequency pulse signals appeared in its electromagnetic spectrum, with amplitudes approaching the system's calibration threshold.

[0176] The node's internal analysis module detects abnormal fluctuations, automatically increasing the sampling frequency to 50Hz and activating a fast caching mechanism. Simultaneously, the node broadcasts an "electromagnetic disturbance coordination signal" to adjacent nodes under the vehicle, triggering the surrounding nodes to increase their frequency from 5Hz to 20Hz, forming a local enhanced perception array for the charging terminal area.

[0177] Using high-frequency sampling, the system successfully captured multiple repetitive pulse waveforms, confirming improper modulation behavior in the radar module's power controller. Testers promptly intervened and adjusted the module's parameters, preventing a potential electromagnetic compatibility failure.

[0178] See also Figure 2 In another embodiment, the present invention further provides an electric vehicle electromagnetic radiation monitoring system, comprising:

[0179] Multiple electromagnetic radiation sensing nodes are arranged at multiple preset locations inside and outside the electric vehicle, and each sensing node is used to collect the time domain signal and frequency domain characteristics of electromagnetic radiation at the corresponding location;

[0180] The electromagnetic cognitive modeling module is used to integrate multi-dimensional input variables including vehicle operating parameters, on-board electronic module status, occupant distribution information, and external environmental interference background to establish a graphical model describing the evolution of electromagnetic radiation distribution;

[0181] A coupling analysis module, provided in the electromagnetic cognitive modeling system, is used to construct a coupling model between multiple electromagnetic source modules within the electric vehicle and establish an electromagnetic induced causal association diagram to identify potential nonlinear radiation enhancement events caused by module synergy or resonance effects, and introduce an induced risk score as a correction factor into the diagram model;

[0182] A priority assessment module is used to predict the high radiation areas that may appear in the electric vehicle under the current operating conditions based on the graph model, and determine the dynamic sampling priority of each sensor node. The priority is determined by the electromagnetic change rate of the node area, the spatial radiation gradient, and the risk level of coupling induced between modules;

[0183] The frequency adjustment and collaborative sampling module is used to autonomously adjust the sampling frequency according to the sampling priority of each sensor node and the local real-time monitoring results, and to achieve local enhanced sampling of predicted high-risk areas through an inter-node collaboration mechanism.

[0184] In a further reality, electromagnetic radiation sensing nodes are set in the battery pack area, the drive motor area, the periphery of the power electronic module, the vehicle communication system location, the charging interface area, the key locations of the passenger compartment, and the external reference points of the vehicle body.

[0185] In a further reality, the nodes in the graph model include state nodes, space nodes, module nodes and risk nodes. The graph model is trained through a graph neural network and outputs electromagnetic radiation intensity prediction values ​​and risk scores based on the temporal evolution information and structural connection relationships of each node.

[0186] In a further reality, the inter-module coupling model includes: an electrical direct coupling path, a spatial field coupling path, and a frequency modulation coupling path. The coupling path is represented in the form of a graph structure, where the edge weight is used to characterize the coupling strength or resonance probability between modules and participates in the construction of an electromagnetically induced causal relationship graph.

[0187] In a further reality, each sensor node includes a local trend detection module and a collaborative perception module. The local trend detection module is used to determine whether to up-sample based on the electromagnetic signal fluctuation within a short time window, and the collaborative perception module is used to broadcast the abnormal status to neighboring nodes to achieve multi-node joint response sampling in high-risk areas.

[0188] It should be noted that the explanation of the above-mentioned electric vehicle electromagnetic radiation monitoring method embodiment is also applicable to the device of the embodiment of the present application and will not be repeated here.

[0189] Those skilled in the art will appreciate that the various units and algorithm steps described in the embodiments disclosed herein can be implemented using a combination of electronic hardware, computer software, and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0190] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0191] In the several embodiments provided in this application, if any function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of this application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory; hereinafter referred to as: ROM), random access memory (Random Access Memory; hereinafter referred to as: RAM), magnetic disk or optical disk, and other media that can store program code.

[0192] The above is only a specific embodiment of the present application. Any person skilled in the art can easily think of changes or replacements within the technical scope disclosed in this application, which should be included in the scope of protection of this application. For some module structures that are not particularly clear in the present invention, the content recorded in the prior art shall prevail. The prior art mentioned in the above background technology section and the specific embodiment section of the present invention can be regarded as part of the present invention and is used to understand the meaning of some technical features or parameters.

Claims

1. A method for monitoring electromagnetic radiation of an electric vehicle, characterized in that: The steps include: Electromagnetic radiation sensor nodes are deployed at multiple preset locations inside and outside the electric vehicle, and each sensor node is used to collect the time domain signal and frequency domain characteristics of electromagnetic radiation at the corresponding location; Establishing a multi-source input electromagnetic cognitive modeling system that integrates multi-dimensional input variables, including vehicle operating parameters, onboard electronic module status, occupant distribution information, and external environmental interference background, to generate a graphical model that describes the evolution of electromagnetic radiation distribution. A coupling model between multiple electromagnetic source modules within an electric vehicle is constructed in the electromagnetic cognitive modeling system, and an electromagnetic induced causal association diagram is established to identify potential nonlinear radiation enhancement events caused by module synergy or resonance effects, and an induced risk score is introduced into the diagram model as a correction factor; Based on the graph model, the high radiation areas that may occur in the electric vehicle under the current operating conditions are predicted, and the dynamic sampling priority of each sensor node is determined. The priority is determined by the electromagnetic change rate, spatial radiation gradient and inter-module coupling-induced risk level of the node area. According to the sampling priority of each sensor node and the local real-time monitoring results, the sampling frequency is adjusted autonomously, and local enhanced sampling of predicted high-risk areas is achieved through the inter-node collaboration mechanism.

2. The electric vehicle electromagnetic radiation monitoring method according to claim 1, characterized in that: The sensor nodes are arranged in the battery pack area, the drive motor area, the periphery of the power electronic module, the vehicle communication system position, the charging interface area, the key positions of the passenger compartment and the external reference points of the vehicle body.

3. The electric vehicle electromagnetic radiation monitoring method according to claim 1, characterized in that: The nodes in the graph model include state nodes, space nodes, module nodes and risk nodes. The graph model is trained using a graph neural network and outputs electromagnetic radiation intensity prediction values ​​and risk scores based on the temporal evolution information and structural connection relationships of each node.

4. The electric vehicle electromagnetic radiation monitoring method according to claim 1, characterized in that: The coupling model includes: an electrical direct coupling path, a spatial field coupling path, and a frequency modulation coupling path. The electrical direct coupling path, the spatial field coupling path, and the frequency modulation coupling path are represented in the form of a graph structure, where the edge weights are used to characterize the coupling strength or resonance probability between modules and participate in the construction of an electromagnetically induced causal association graph.

5. The electric vehicle electromagnetic radiation monitoring method according to claim 1, characterized in that: Each sensor node includes a local trend detection module and a collaborative perception module. The local trend detection module is used to determine whether to up-sample based on the electromagnetic signal fluctuations within a short time window. The collaborative perception module is used to broadcast abnormal conditions to neighboring nodes to achieve multi-node joint response sampling in high-risk areas.

6. An electric vehicle electromagnetic radiation monitoring system, characterized in that: The system includes the following modules: Multiple electromagnetic radiation sensing nodes are arranged at multiple preset locations inside and outside the electric vehicle, and each sensing node is used to collect the time domain signal and frequency domain characteristics of electromagnetic radiation at the corresponding location; The electromagnetic cognitive modeling module is used to integrate multi-dimensional input variables including vehicle operating parameters, on-board electronic module status, occupant distribution information, and external environmental interference background to establish a graphical model describing the evolution of electromagnetic radiation distribution; A coupling analysis module, provided in the electromagnetic cognitive modeling module, is used to construct a coupling model between multiple electromagnetic source modules within the electric vehicle and establish an electromagnetic induced causal association diagram to identify potential nonlinear radiation enhancement events caused by module synergy or resonance effects, and introduce the induced risk score as a correction factor into the diagram model; A priority assessment module is used to predict the high radiation areas that may appear in the electric vehicle under the current operating conditions based on the graph model, and determine the dynamic sampling priority of each sensor node. The priority is determined by the electromagnetic change rate of the node area, the spatial radiation gradient, and the risk level of coupling induced between modules; The frequency adjustment and collaborative sampling module is used to autonomously adjust the sampling frequency according to the sampling priority of each sensor node and the local real-time monitoring results, and to achieve local enhanced sampling of predicted high-risk areas through an inter-node collaboration mechanism.

7. The electric vehicle electromagnetic radiation monitoring system according to claim 6, characterized in that: The electromagnetic radiation sensing nodes are arranged in the battery pack area, the drive motor area, the periphery of the power electronic module, the vehicle communication system position, the charging interface area, the key positions of the passenger compartment and the external reference points of the vehicle body.

8. The electric vehicle electromagnetic radiation monitoring system according to claim 6, characterized in that: The nodes in the graph model include state nodes, space nodes, module nodes and risk nodes. The graph model is trained through a graph neural network and outputs electromagnetic radiation intensity prediction values ​​and risk scores based on the temporal evolution information and structural connection relationship of each node.

9. The electric vehicle electromagnetic radiation monitoring system according to claim 6, characterized in that: The coupling model includes: an electrical direct coupling path, a spatial field coupling path, and a frequency modulation coupling path. The electrical direct coupling path, the spatial field coupling path, and the frequency modulation coupling path are represented in the form of a graph structure, where the edge weights are used to characterize the coupling strength or resonance probability between modules and participate in the construction of an electromagnetically induced causal association graph.

10. The electric vehicle electromagnetic radiation monitoring system according to claim 6, characterized in that: Each sensor node includes a local trend detection module and a collaborative perception module. The local trend detection module is used to determine whether to up-sample based on the electromagnetic signal fluctuations within a short time window. The collaborative perception module is used to broadcast abnormal conditions to neighboring nodes to achieve multi-node joint response sampling in high-risk areas.

Images