An electric vehicle driving state sensing method and system

Abstract

The application is suitable for the field of intelligent perception of electric vehicles, and provides an electric vehicle driving state perception method and system.The method comprises the following steps: continuously monitoring the electromagnetic compatibility background noise of the vehicle ECU, collecting electromagnetic interference signals and converting them into digitized frequency spectrum data; performing time-frequency analysis on the frequency spectrum data, extracting feature parameters to generate a feature matrix, and combining a typical road condition sample library to train and establish a road condition fingerprint classification model; inputting the real-time electromagnetic interference feature matrix into the model, identifying the road surface type and outputting an adhesion coefficient estimate; combining vehicle driving parameters to calculate safety parameters, predicting trajectory stability boundaries and generating evaluation results; monitoring abnormal changes in electromagnetic interference, combining vehicle parameters to judge the slip risk level and triggering a graded warning.The scheme does not require additional sensors, improves the perception accuracy and timeliness, and ensures the driving safety of electric vehicles.

Description

Technical Field

[0001] This invention belongs to the field of intelligent sensing for electric vehicles, and particularly relates to a method and system for sensing the driving status of electric vehicles. Background Technology

[0002] With the rapid development of the electric vehicle industry, its driving safety and intelligence level have become key concerns. Driving status perception, as a core component ensuring driving safety, needs to accurately identify and warn of road surface types, driving stability, and the risk of skidding. Currently, the industry relies heavily on traditional mechanical or visual sensors for driving status perception, collecting vehicle dynamic parameters or road visual information to achieve basic perception. Meanwhile, the application of electromagnetic compatibility technology in electric vehicle electronic control units (ECUs) is becoming increasingly widespread. The electromagnetic interference signals generated during ECU operation have a potential correlation with vehicle driving status and road excitation, providing a new technical path for driving status perception. The industry is gradually exploring sensing solutions based on electromagnetic signals to improve perception accuracy and adaptability.

[0003] Among the closest existing technologies, some electric vehicle driving status perception solutions rely on mechanical parameters such as wheel speed and acceleration. However, this method has low sensitivity to subtle changes in the road surface, and there is a lag in capturing signs of slippage, making it difficult to provide early warnings. On the other hand, vision-based solutions are easily affected by severe weather or insufficient lighting, and the accuracy of road surface recognition is unstable. This results in insufficient accuracy, real-time performance, and reliability of driving status perception, making it difficult to meet the driving safety requirements of electric vehicles under complex road conditions. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for sensing the driving status of an electric vehicle, aiming to solve the technical problems existing in the prior art as identified in the background art.

[0005] This invention is implemented as follows: a method for sensing the driving state of an electric vehicle, the method comprising:

[0006] Continuously monitor the electromagnetic compatibility background noise of the vehicle's electronic control unit, collect electromagnetic interference signals, and convert them into digital electromagnetic interference noise spectrum data;

[0007] Time-frequency analysis is performed on the electromagnetic interference noise spectrum data to extract a set of feature parameters and generate a feature matrix. The feature matrix is ​​then matched and trained with a typical road condition sample library to establish a road condition fingerprint classification model.

[0008] The electromagnetic interference feature matrix acquired in real time is input into the road condition fingerprint classification model to calculate the similarity between the real-time road condition and the standard road condition fingerprints in the road condition fingerprint classification model, identify the current road surface type, and output the corresponding road surface adhesion coefficient estimate.

[0009] Based on the identified current road surface type and estimated road surface adhesion coefficient, combined with the vehicle's current driving state parameters, the vehicle safety parameters are calculated, the stability boundary of the driving trajectory is predicted, and the trajectory stability assessment result is generated.

[0010] The system monitors the dynamic changes of electromagnetic interference noise in the preset frequency band in real time, identifies whether there is spectrum diffusion and specific harmonic energy abrupt change, and if so, determines the risk level of wheel slippage by combining the vehicle's current driving status parameters and triggers a graded warning signal.

[0011] As a further embodiment of the present invention, the acquisition of electromagnetic interference signals specifically includes:

[0012] Simultaneously collect conducted and radiated emission signals generated by various electronic control units of the vehicle in the 150kHz to 1GHz frequency band to obtain the original electromagnetic interference signals;

[0013] The original electromagnetic interference signal is preprocessed and acquired, and the preprocessed original electromagnetic interference signal is converted into digital electromagnetic interference noise spectrum data.

[0014] As a further aspect of the present invention, the establishment of the road condition fingerprint classification model specifically includes:

[0015] For the electromagnetic interference noise spectrum data, the power spectral density feature vector is extracted as a frequency domain feature, and the peak factor, impulse factor and kurtosis index of the signal are calculated as time domain statistical features.

[0016] The power spectral density feature vector, harmonic component distribution and time-domain statistical features are combined to form an initial feature matrix. The initial feature matrix is ​​then dimensionality-reduced to obtain the dimensionality-reduced feature matrix.

[0017] The feature matrix is ​​trained using unsupervised learning with standard road condition samples from a typical road condition sample library to establish a road condition fingerprint classification model that includes five standard road conditions: asphalt pavement, concrete pavement, gravel pavement, snow pavement, and icy pavement.

[0018] As a further aspect of the present invention, the step of identifying the current road surface type and outputting the corresponding road surface adhesion coefficient estimate specifically includes:

[0019] The system collects the vehicle's current electromagnetic interference signals in real time and converts them into a real-time electromagnetic interference feature matrix.

[0020] Input the real-time electromagnetic interference feature matrix into the road condition fingerprint classification model, and calculate the similarity value between the real-time road condition features and the standard road condition fingerprint features in the road condition fingerprint classification model.

[0021] Compare the similarity values ​​and select the road surface type corresponding to the standard road condition fingerprint with the largest similarity value as the current road surface type identification result;

[0022] Based on the current road surface type identification result, query the pre-stored road surface type and adhesion coefficient mapping table, and output the corresponding road surface adhesion coefficient estimate.

[0023] As a further aspect of the present invention, the generated trajectory stability evaluation result specifically includes:

[0024] The vehicle's current speed, steering angle, and acceleration driving status parameters are obtained. Combined with the identified current road surface type and road adhesion coefficient estimate, the maximum safe speed, minimum braking distance, and critical steering angle are calculated.

[0025] The vehicle stability boundary envelope is constructed using the maximum safe speed, minimum braking distance, and critical steering angle. The stability boundary of the driving trajectory is predicted by comparing the relative position of the current driving state with the vehicle stability boundary envelope in real time.

[0026] When the current driving state is detected to be in line with or exceeds the stability boundary of the driving trajectory, a trajectory stability assessment result containing the stability level and adjustment suggestions is generated.

[0027] As a further aspect of the present invention, the identification of whether there is spectral diffusion and specific harmonic energy abrupt change specifically includes:

[0028] Within a preset frequency band, the spectral spread width and third harmonic energy change rate of electromagnetic interference noise are monitored in real time. When the spectral spread width exceeds the reference value of 100kHz and the third harmonic energy change rate exceeds 40dB / ms, it is determined that there are abnormal spectral characteristics.

[0029] When abnormal spectral characteristics are detected, the vehicle’s current wheel speed signal and yaw rate are obtained, and the wheel slip risk level is calculated.

[0030] A warning signal is triggered based on the wheel slippage risk level.

[0031] Another object of the present invention is to provide an electric vehicle driving state perception system, the system comprising:

[0032] The data acquisition module is used to continuously monitor the electromagnetic compatibility background noise of the vehicle's electronic control unit, collect electromagnetic interference signals, and convert them into digital electromagnetic interference noise spectrum data.

[0033] The time-frequency analysis and feature extraction module is used to perform time-frequency analysis on the electromagnetic interference noise spectrum data, extract a set of feature parameters, generate a feature matrix, and match and train the feature matrix with a typical road condition sample library to establish a road condition fingerprint classification model.

[0034] The real-time road condition recognition module is used to input the electromagnetic interference feature matrix acquired in real time into the road condition fingerprint classification model, calculate the similarity between the real-time road condition and various standard road condition fingerprints in the road condition fingerprint classification model, identify the current road surface type, and output the corresponding road surface adhesion coefficient estimate.

[0035] The trajectory stability assessment module is used to calculate vehicle safety parameters, predict the stability boundary of the driving trajectory, and generate trajectory stability assessment results based on the identified current road surface type and road surface adhesion coefficient estimate, combined with the vehicle's current driving state parameters.

[0036] The risk assessment module is used to monitor the dynamic changes of electromagnetic interference noise in the preset frequency band in real time, identify whether there is a spectrum diffusion phenomenon and a sudden change in specific harmonic energy. If so, it determines the risk level of wheel slippage by combining the current driving status parameters of the vehicle and triggers a graded warning signal.

[0037] As a further embodiment of the present invention, the real-time traffic recognition module includes:

[0038] The feature matrix generation module is used to collect the current electromagnetic interference signal of the vehicle in real time and convert it into a real-time electromagnetic interference feature matrix.

[0039] The road condition fingerprint classification model input module is used to input the real-time electromagnetic interference feature matrix into the road condition fingerprint classification model and calculate the similarity value between the real-time road condition features and various standard road condition fingerprint features in the road condition fingerprint classification model.

[0040] The road surface type recognition module is used to compare the similarity values ​​and select the road surface type corresponding to the standard road condition fingerprint with the largest similarity value as the current road surface type recognition result.

[0041] The adhesion coefficient query module is used to query the pre-stored mapping table of road surface type and adhesion coefficient based on the current road surface type identification result, and output the corresponding road surface adhesion coefficient estimate.

[0042] As a further embodiment of the present invention, the trajectory stability evaluation module includes:

[0043] The safety index calculation module is used to obtain the vehicle's current speed, steering angle and acceleration driving status parameters, and combine them with the identified current road surface type and road surface adhesion coefficient estimate to calculate the maximum safe speed, minimum braking distance and critical steering angle.

[0044] The vehicle stability boundary envelope construction module is used to construct the vehicle stability boundary envelope using the maximum safe vehicle speed, minimum braking distance and critical steering angle, and predict the stability boundary of the driving trajectory by comparing the relative position of the current driving state with the vehicle stability boundary envelope in real time.

[0045] The trajectory evaluation module is used to generate a trajectory stability evaluation result that includes a stability level and adjustment suggestions when the current driving state is detected to be in line with or exceeds the stability boundary of the driving trajectory.

[0046] As a further embodiment of the present invention, the risk assessment module includes:

[0047] The anomaly detection module is used to monitor the spectral spread width and third harmonic energy change rate of electromagnetic interference noise in real time within a preset frequency band. When the spectral spread width exceeds the reference value of 100kHz and the third harmonic energy change rate exceeds 40dB / ms, it is determined that there are abnormal spectral characteristics.

[0048] The wheel slip risk level calculation module is used to obtain the vehicle's current wheel speed signal and yaw rate when abnormal spectral characteristics are detected, and to calculate the wheel slip risk level.

[0049] The warning triggering module is used to trigger a warning signal based on the wheel slippage risk level.

[0050] The beneficial effects of this invention are:

[0051] This solution continuously monitors the electromagnetic compatibility background noise of the electric vehicle's ECU without requiring additional dedicated sensors, fully utilizing existing electronic system resources and reducing hardware costs and system complexity. By extracting multi-dimensional features of electromagnetic interference signals through time-frequency analysis, a fingerprint classification model encompassing five standard road conditions is constructed. Combined with unsupervised learning, this improves road condition recognition accuracy, effectively avoiding the environmental impact issues of traditional vision-based solutions and enhancing adaptability in complex environments.

[0052] By correlating road condition recognition results with the road surface adhesion coefficient and combining them with real-time vehicle driving parameters to calculate safety parameters and construct a stable boundary envelope, a quantitative assessment of driving trajectory stability is achieved. Simultaneously, abnormal changes in electromagnetic interference are monitored, and parameters such as wheel speed and yaw rate are integrated to determine the risk of slippage and trigger graded warnings, thus capturing early signs of slippage in advance. This solves the problem of lag in traditional mechanical parameter perception and comprehensively improves the accuracy, real-time performance, and reliability of electric vehicle driving status perception, providing strong protection for driving safety in complex road conditions. Attached Figure Description

[0053] Figure 1 A flowchart of an electric vehicle driving state perception method provided in an embodiment of the present invention;

[0054] Figure 2 A flowchart for collecting electromagnetic interference signals provided in an embodiment of the present invention;

[0055] Figure 3 This is a flowchart of establishing a road condition fingerprint classification model provided in an embodiment of the present invention;

[0056] Figure 4 A flowchart for identifying the current road surface type provided in an embodiment of the present invention;

[0057] Figure 5 A flowchart for generating trajectory stability evaluation results provided in an embodiment of the present invention;

[0058] Figure 6 A flowchart for identifying the presence of spectral diffusion and specific harmonic energy abrupt changes is provided for embodiments of the present invention;

[0059] Figure 7 A structural block diagram of an electric vehicle driving status perception system provided in an embodiment of the present invention;

[0060] Figure 8 This is a structural block diagram of the real-time traffic recognition module provided in an embodiment of the present invention;

[0061] Figure 9 This is a structural block diagram of the trajectory stability evaluation module provided in an embodiment of the present invention;

[0062] Figure 10 This is a structural block diagram of the risk assessment module provided in an embodiment of the present invention. Detailed Implementation

[0063] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0064] Figure 1 A flowchart of an electric vehicle driving state perception method provided in an embodiment of the present invention is shown below. Figure 1 As shown, the method includes:

[0065] S100 continuously monitors the electromagnetic compatibility background noise of the vehicle's electronic control unit, collects electromagnetic interference signals, and converts them into digital electromagnetic interference noise spectrum data.

[0066] The system synchronously collects conducted and radiated emission signals generated by each electronic control unit within a set frequency band. The conducted emission signals originate from the internal circuit conduction path of the electronic equipment, while the radiated emission signals originate from the electromagnetic signals radiated outward by the electronic equipment. The wide-band coverage design can capture the full-band characteristics of electromagnetic interference under different road conditions.

[0067] The collected raw electromagnetic interference signals undergo preprocessing operations such as filtering, amplification, and noise reduction to filter out electromagnetic interference from the external environment that is unrelated to the vehicle's driving state and enhance the effective signal components caused by road excitation and vehicle vibration. Then, the preprocessed analog signals are converted into digital electromagnetic interference noise spectrum data by high-speed data acquisition equipment. The data format is adapted to the processing requirements of subsequent time-frequency analysis algorithms to ensure that the data can accurately reflect the amplitude changes, frequency distribution, and temporal evolution of electromagnetic interference.

[0068] S200, perform time-frequency analysis on the electromagnetic interference noise spectrum data, extract the set of feature parameters, and generate a feature matrix. Then, match and train the feature matrix with a typical road condition sample library to establish a road condition fingerprint classification model.

[0069] Time-domain analysis focuses on the dynamic fluctuation patterns of signals, capturing the instantaneous impact characteristics of electromagnetic interference by calculating the peak factor, impulse factor, and kurtosis index of the signal. These indicators can reflect the differences in the impact intensity of vehicle vibration on the electromagnetic signals of electronic control units under different road conditions. Frequency-domain analysis focuses on exploring the frequency distribution and energy distribution patterns of signals, extracting power spectral density feature vectors to characterize the proportion of electromagnetic energy in different frequency bands, and analyzing the distribution of harmonic components to clarify the energy proportion and frequency position of each harmonic in the electromagnetic signal. Under different road conditions, the energy concentration frequency band and harmonic composition of electromagnetic interference are significantly different. For example, when driving on asphalt roads, the power spectral density of electromagnetic interference is mostly concentrated in the low frequency band and the harmonic components are simple, while when driving on gravel roads, the power spectral density shows more energy peaks in the mid-to-high frequency band and the harmonic order increases.

[0070] The extracted power spectral density feature vectors, harmonic component distributions, and time-domain statistical features are integrated according to sample dimensions to construct an initial feature matrix. Each matrix row vector corresponds to a complete set of electromagnetic interference features at a given driving time, and each matrix column vector corresponds to a feature parameter, achieving systematic integration of multi-dimensional features. The initial feature matrix is ​​then subjected to dimensionality reduction processing to eliminate redundant information and correlation interference between features, retaining the core features most discriminative for road condition identification, forming a simplified feature matrix.

[0071] The dimensionality-reduced feature matrix is ​​trained using unsupervised learning with standard data from a typical road condition sample library. The sample library contains electromagnetic interference feature samples from a large number of driving scenarios under five standard road conditions. During the training process, the model learns and solidifies the electromagnetic interference feature patterns corresponding to each standard road condition through automatic clustering and feature matching. Finally, a road condition fingerprint classification model that can classify road conditions is established. This model can output the corresponding road condition category based on the input electromagnetic interference feature matrix.

[0072] In actual training, driving data for five standard road conditions under different weather conditions can be covered, such as asphalt roads in dry and wet environments, snow roads with different snow thicknesses, and icy roads with different degrees of icing. Through training with a large number of samples, the model can adapt to the influence of environmental changes on electromagnetic interference characteristics and improve the model's adaptability to complex real-world scenarios.

[0073] S300, input the real-time acquired electromagnetic interference feature matrix into the road condition fingerprint classification model, calculate the similarity between the real-time road condition and the standard road condition fingerprints in the road condition fingerprint classification model, identify the current road surface type, and output the corresponding road surface adhesion coefficient estimate.

[0074] The collected raw electromagnetic interference signals need to undergo a preprocessing process to remove environmentally irrelevant interference, extract power spectral density feature vectors and time-domain statistical features through time-frequency analysis, and simplify the feature dimensions through dimensionality reduction processing to finally form a real-time electromagnetic interference feature matrix. The dimensions and data format of this matrix are completely matched with the feature matrix used in training the road condition fingerprint classification model in S200, ensuring that the data input to the model can be effectively identified and calculated, and avoiding identification deviations caused by differences in feature formats.

[0075] After inputting the real-time electromagnetic interference feature matrix into the established road condition fingerprint classification model, the model calls upon the fingerprint feature vectors of five standard road conditions—asphalt pavement, concrete pavement, gravel pavement, snow-covered pavement, and icy pavement—stored internally. It then compares the real-time features with each standard feature using a similarity calculation method to uncover the correlation between the real-time electromagnetic signal features and known road condition features. Different road conditions result in varying vehicle vibration intensity and frequency, leading to inherent differences in the electromagnetic interference characteristics emitted by the electronic control unit. By calculating the similarity score, the standard road condition type that best matches the current driving environment can be identified.

[0076] After determining the current road surface type, the system will call the pre-stored road surface type and adhesion coefficient mapping table. This mapping table is built based on a large amount of driving test data under different road conditions, covering the typical adhesion coefficient range of various road surfaces under different environmental conditions such as dry, wet, and low temperature. It can quickly output the corresponding road surface adhesion coefficient estimate based on the identified road surface type. For example, the adhesion coefficient of asphalt road surface in a wet environment will be lower than that of the same road surface type in a dry environment, and the adhesion coefficient of snow-covered road surface will be significantly lower than that of concrete road surface.

[0077] In real-world driving scenarios, if a vehicle drives from an asphalt road onto a snow-covered road, the electromagnetic interference signal features collected in real time will gradually converge towards the standard fingerprint of the snow-covered road. In the similarity calculation results, the similarity value of the snow-covered road will gradually increase and exceed that of other types. The system will then update the road type identification results and output the corresponding low adhesion coefficient estimate.

[0078] S400 calculates vehicle safety parameters, predicts the stability boundary of the driving trajectory, and generates trajectory stability assessment results based on the identified current road surface type and estimated road surface adhesion coefficient, combined with the vehicle's current driving state parameters.

[0079] The acquired real-time driving status parameters are correlated and integrated with the current road surface type and the corresponding estimated road surface adhesion coefficient to calculate vehicle safety parameters. When calculating the maximum safe speed, the upper limit of the road surface lateral grip force determined by the road surface adhesion coefficient is used, combined with the real-time turning radius, to determine the highest driving speed at which the vehicle will not skid when turning on the current road surface. When calculating the minimum braking distance, the shortest safe distance required for the vehicle to stop is derived from the current speed based on the maximum braking deceleration of the road surface corresponding to the adhesion coefficient, avoiding collisions caused by insufficient braking distance. When calculating the critical steering angle, the maximum steering angle of the vehicle without loss of control such as oversteer or understeer is determined by combining the vehicle's fixed wheelbase and the current speed.

[0080] After calculating the three safety parameters, a two-dimensional coordinate system is established with vehicle speed as the horizontal axis and steering angle as the vertical axis. The maximum safe vehicle speed corresponding to different steering radii is plotted as the maximum safe vehicle speed-steering angle curve. The minimum braking distance corresponding to different vehicle speeds is converted into the braking safety boundary curve. The critical steering angle is plotted as the horizontal steering angle upper limit curve. The three curves together form a closed vehicle stability boundary envelope. The area inside the envelope is the safe driving range under the current road conditions.

[0081] By comparing the relative positions of the current vehicle speed, steering angle, and stability boundary envelope in real time, and combining the stability margin to quantify the distance between the current driving state and the stability boundary, the stability margin comprehensively considers the degree to which the vehicle speed exceeds the maximum safe speed, the extent to which the steering angle approaches the critical steering angle, and the difference between the actual distance to the vehicle and the minimum braking distance. The magnitude of the value directly reflects the stability level of the driving state.

[0082] When the system detects that the current driving state is in line with or exceeds the stability boundary envelope, it generates a corresponding stability level based on the specific range of the stability margin, and simultaneously matches targeted adjustment suggestions. For example, when driving on an icy road, if the current speed exceeds the maximum safe speed and the steering angle is close to the critical value, the stability margin will be negative, and the system will generate an instability level assessment result, and provide suggestions to trigger the ESP system, adjust torque distribution, and prompt the driver to decelerate urgently. If driving on an asphalt road, if the current speed is below the maximum safe speed and the steering is smooth, the stability margin will be in a higher range, and a high stability level assessment result will be generated, suggesting that the current driving state be maintained.

[0083] The S500 monitors the dynamic changes of electromagnetic interference noise in a preset frequency band in real time, identifies whether there is a spectrum diffusion phenomenon and a sudden change in specific harmonic energy. If so, it determines the risk level of wheel slippage by combining the vehicle's current driving status parameters and triggers a graded warning signal.

[0084] The monitoring focuses on two key indicators: spectral spread width and third harmonic energy change rate. Spectral spread width reflects the change in the distribution range of electromagnetic interference signals on the frequency axis, while third harmonic energy change rate reflects the instantaneous fluctuation amplitude of electromagnetic energy in a specific frequency band.

[0085] When a vehicle travels on different road conditions or experiences signs of slippage, changes in the road surface friction coefficient lead to alterations in the vehicle's vibration intensity and frequency. This, in turn, causes changes in the contact state of the internal circuits and the coupling relationships of the wires within the electronic control unit, causing electromagnetic interference signals to spread from their originally concentrated frequency range to a wider frequency band. Simultaneously, abnormal vibrations in transmission components such as the motor and reducer exacerbate fluctuations in third harmonic energy. By tracking these two indicators in real time, when the spectral spread width and the rate of change of third harmonic energy exceed the reference values, abnormal spectral characteristics can be identified. This anomaly is an early signal that the vehicle's driving state is about to become unbalanced.

[0086] Once an anomaly is detected, the vehicle's current wheel speed signal and yaw rate are immediately acquired. The wheel speed signal is used to calculate the difference in rotational speed between the left and right wheels, reflecting whether the wheels are slipping or have a tendency to slide. The yaw rate is used to determine whether the vehicle body has deviated from the expected trajectory. The abnormal spectrum characteristics are fused with the wheel speed difference and yaw rate data to calculate the wheel slip risk level.

[0087] Based on the risk level, corresponding warning signals are triggered. At a low risk level, warning information is transmitted through the in-vehicle prompt device to remind the driver to adjust the operation. At a medium risk level, the vehicle stability control system is activated to initially intervene and slightly adjust the power output. At a high risk level, the stability control system is activated to deeply intervene and directly correct the driving state by adjusting torque distribution and intervening in braking.

[0088] like Figure 2 As shown, the acquisition of electromagnetic interference signals specifically includes:

[0089] S110 synchronously collects conducted and radiated emission signals generated by various electronic control units of the vehicle in the 150kHz to 1GHz frequency band to obtain the original electromagnetic interference signal;

[0090] S120, the original electromagnetic interference signal is preprocessed and the signal is acquired, and the preprocessed original electromagnetic interference signal is converted into digital electromagnetic interference noise spectrum data.

[0091] like Figure 3 As shown, the establishment of the road condition fingerprint classification model specifically includes:

[0092] S210, For the electromagnetic interference noise spectrum data, extract the power spectral density feature vector as a frequency domain feature, and simultaneously calculate the signal's peak factor, impulse factor, and kurtosis index as time domain statistical features.

[0093] S220, the power spectral density feature vector, harmonic component distribution and time-domain statistical features are combined to form an initial feature matrix, and the initial feature matrix is ​​reduced in dimension to obtain the reduced feature matrix;

[0094] The initial characteristic matrix is ​​composed of the power spectral density eigenvector, harmonic component distribution, and time-domain statistical characteristics, and is denoted as the initial characteristic matrix. ,in This refers to the number of samples (i.e., the number of EMI data samples collected at different driving times). The feature dimensions are (power spectral density feature vector dimension + harmonic component distribution dimension + time-domain statistical feature dimension).

[0095] Principal component analysis (PCA) was used for dimensionality reduction, and the resulting feature matrix is: ( , (The feature dimension after dimensionality reduction), the calculation formula is:

[0096] ;

[0097] in:

[0098] The initial feature matrix, where each row represents the complete features of a sample and the columns correspond to specific features;

[0099] The PCA transformation matrix is ​​derived from the covariance matrix of the initial eigenma matrix. The feature vectors corresponding to the largest eigenvalues ​​are used to extract the main features and reduce the dimensionality.

[0100] The reduced-dimensional feature matrix retains the most discriminative information from the initial features and is used for subsequent training of the road condition fingerprint classification model.

[0101] S230, The feature matrix is ​​trained by unsupervised learning with standard road condition samples in the typical road condition sample library to establish a road condition fingerprint classification model that includes five standard road conditions: asphalt pavement, concrete pavement, gravel pavement, snow pavement and icy pavement.

[0102] like Figure 4 As shown, the step of identifying the current road surface type and outputting the corresponding road surface adhesion coefficient estimate specifically includes:

[0103] S310 collects the vehicle's current electromagnetic interference signals in real time and converts them into a real-time electromagnetic interference feature matrix.

[0104] S320, Input the real-time electromagnetic interference feature matrix into the road condition fingerprint classification model, and calculate the similarity value between the real-time road condition features and the various standard road condition fingerprint features in the road condition fingerprint classification model.

[0105] The cosine similarity is used to calculate the similarity between real-time traffic features and standard traffic fingerprint features. The formula is as follows:

[0106] ;

[0107] in:

[0108] Real-time traffic characteristics and the first Similarity value of fingerprint features of standard road conditions ( =1,2,3,4,5 (corresponding to asphalt, concrete, gravel, snow, and icy road surfaces), with a value range of [0,1]. The larger the value, the higher the similarity.

[0109] : The eigenvectors of the real-time electromagnetic interference feature matrix after dimensionality reduction;

[0110] The typical road condition sample library after dimensionality reduction Fingerprint feature vectors for standard road conditions.

[0111] S330, compare the similarity values ​​and select the road surface type corresponding to the standard road condition fingerprint with the largest similarity value as the current road surface type identification result;

[0112] S340, Based on the current road surface type identification result, query the pre-stored road surface type and adhesion coefficient mapping table, and output the corresponding road surface adhesion coefficient estimate.

[0113] like Figure 5 As shown, the generated trajectory stability evaluation results specifically include:

[0114] S410: Obtain the vehicle's current speed, steering angle, and acceleration driving status parameters, and combine them with the identified current road surface type and road surface adhesion coefficient estimate to calculate the maximum safe speed, minimum braking distance, and critical steering angle.

[0115] (1) Maximum safe speed :

[0116] Based on the lateral force balance of a vehicle while driving on a curve, the maximum safe speed is limited by the road surface adhesion coefficient and the turning radius to avoid skidding. The calculation formula is as follows:

[0117] ;

[0118] in:

[0119] : The maximum safe speed under current road conditions;

[0120] The estimated value of the road surface adhesion coefficient is obtained by querying the mapping table based on the road condition identification results;

[0121] Gravitational acceleration;

[0122] Real-time turning radius, calculated from steering angle and vehicle wheelbase.

[0123] (2) Minimum braking distance :

[0124] Based on braking dynamics, and under the constraint of road adhesion coefficient, the shortest distance for a vehicle to stop from its current speed v_0, neglecting braking system response delay, is calculated using the following formula:

[0125] ;

[0126] in:

[0127] Minimum braking distance under current road conditions;

[0128] : The vehicle's current speed.

[0129] (3) Critical steering angle :

[0130] Based on the vehicle steady-state steering model, the critical steering angle is the steering angle at which the vehicle is about to sideslip, and it is related to vehicle speed, wheelbase, and coefficient of adhesion. The calculation formula is as follows:

[0131] ;

[0132] in:

[0133] Critical steering angle;

[0134] Vehicle wheelbase.

[0135] S420: Construct a vehicle stability boundary envelope using the maximum safe speed, minimum braking distance, and critical steering angle; predict the stability boundary of the driving trajectory by comparing the relative position of the current driving state with the vehicle stability boundary envelope in real time.

[0136] For the vehicle stability boundary envelope:

[0137] by vehicle speed For horizontal axis and steering angle Establish a two-dimensional coordinate system with the vertical axis as the ordinate;

[0138] Based on different turning radii Calculate the corresponding Plot the maximum safe speed-steering angle curve in the coordinate system;

[0139] Based on current vehicle speed calculate Combined with braking deceleration limits, draw the minimum braking distance-vehicle speed auxiliary boundary;

[0140] Plotting the critical steering angle The corresponding horizontal line serves as the upper limit boundary of the steering angle;

[0141] The closed region enclosed by the above curves is the vehicle stability boundary envelope (stability region).

[0142] For the stability boundary of the predicted driving trajectory:

[0143] Using stability boundary margins The formula used to quantify the distance between the current state and the stable boundary is:

[0144] ;

[0145] Calculation basis: Through normalization, the margins of vehicle speed, steering angle and braking distance are considered, and the value range is (-∞, 1]. The larger the value, the more stable it is, and the negative value indicates that it exceeds the boundary.

[0146] in:

[0147] Stability boundary margin;

[0148] The vehicle's current steering angle;

[0149] : The current actual distance to the vehicle in front.

[0150] S430 generates a trajectory stability assessment result that includes a stability level and adjustment suggestions when it detects that the current driving state is in line with and exceeds the stability boundary of the driving trajectory.

[0151] Stability level determination criteria:

[0152] Level 1 stability (high stability): The current driving state is far from the stability boundary and there is no potential risk. It is recommended to maintain the current driving state without adjustment.

[0153] Level 2 stability (intermediate stability): Approaching the stability boundary, it is necessary to pay attention to changes in driving conditions. It is recommended to reduce the accelerator pedal opening, maintain a constant speed, and avoid sharp turns.

[0154] Level 3 Warning (Low Stability): The vehicle is about to exceed the stability boundary and there is a slight risk of instability. The motor output torque will be forcibly reduced (the current vehicle speed will be reduced by 10%-20%) to remind the driver to slow down and reduce the steering angle.

[0155] Level 4 Warning (Unstable): The situation has exceeded the stability boundary, posing serious risks such as sideslip and insufficient braking distance. This triggers the ESP system to intervene (braking one side of the wheels), adjust the torque distribution between the front and rear axles (reducing the torque of the drive wheels by more than 30%), and simultaneously issue an audible and visual alarm to prompt the driver to decelerate urgently.

[0156] like Figure 6 As shown, the identification of whether there is spectral diffusion and specific harmonic energy abrupt change specifically includes:

[0157] S510 monitors the spectral spread width and third harmonic energy change rate of electromagnetic interference noise in real time within a preset frequency band. When the spectral spread width exceeds the reference value of 100kHz and the third harmonic energy change rate exceeds 40dB / ms, it is determined that there are abnormal spectral characteristics.

[0158] Under normal road conditions, vehicle vibration is stable, the EMI spectrum of the ECU is concentrated in a specific frequency band, the diffusion width is usually ≤50kHz, and the harmonic energy changes gradually (≤10dB / ms).

[0159] When the wheels are about to slip, the sudden change in the road friction coefficient causes the vehicle to vibrate more, which can lead to poor contact in the internal circuit of the ECU, vibration coupling of wires, etc., causing the EMI spectrum to spread to a wider frequency band (more than 100kHz). At the same time, abnormal vibration of components such as motors and reducers can cause a sharp increase in third harmonic energy (the measured change rate before slippage is often >40dB / ms).

[0160] S520: When abnormal spectral characteristics are detected, the vehicle’s current wheel speed signal and yaw rate are obtained, and the wheel slip risk level is calculated.

[0161] The risk level is obtained by weighting and summing three indicators: abnormal wheel speed difference, abnormal yaw rate, and sudden change in EMI harmonic energy. The value ranges from [0,1], with higher values ​​indicating greater risk.

[0162] ;

[0163] in:

[0164] The difference in rotational speed between the left and right wheels. Maximum permissible wheel speed difference, a fixed value;

[0165] Actual yaw rate The maximum stable yaw rate is calculated from the vehicle dynamics model based on the current vehicle speed.

[0166] The change in energy of the third harmonic. The threshold for third harmonic energy variation is set at 40 dB.

[0167] Weighting coefficient.

[0168] S530, based on the wheel slippage risk level, triggers a warning signal.

[0169] Figure 7 A structural block diagram of an electric vehicle driving state perception system provided in an embodiment of the present invention is shown below. Figure 7 As shown, the system includes:

[0170] The data acquisition module 100 is used to continuously monitor the electromagnetic compatibility background noise of the vehicle's electronic control unit, collect electromagnetic interference signals, and convert them into digital electromagnetic interference noise spectrum data.

[0171] The time-frequency analysis and feature extraction module 200 is used to perform time-frequency analysis on the electromagnetic interference noise spectrum data, extract a set of feature parameters, generate a feature matrix, and match and train the feature matrix with a typical road condition sample library to establish a road condition fingerprint classification model.

[0172] The real-time road condition recognition module 300 is used to input the electromagnetic interference feature matrix acquired in real time into the road condition fingerprint classification model, calculate the similarity between the real-time road condition and various standard road condition fingerprints in the road condition fingerprint classification model, identify the current road surface type, and output the corresponding road surface adhesion coefficient estimate.

[0173] The trajectory stability assessment module 400 is used to calculate vehicle safety parameters, predict the stability boundary of the driving trajectory, and generate trajectory stability assessment results based on the identified current road surface type and road surface adhesion coefficient estimate, combined with the vehicle's current driving state parameters.

[0174] The risk assessment module 500 is used to monitor the dynamic changes of electromagnetic interference noise in a preset frequency band in real time, identify whether there is a spectrum diffusion phenomenon and a sudden change in specific harmonic energy. If so, it determines the risk level of wheel slippage by combining the current driving status parameters of the vehicle and triggers a graded warning signal.

[0175] like Figure 8 As shown, the real-time traffic recognition module 300 includes:

[0176] The feature matrix generation module 310 is used to collect the current electromagnetic interference signal of the vehicle in real time and convert it into a real-time electromagnetic interference feature matrix.

[0177] The road condition fingerprint classification model input module 320 is used to input the real-time electromagnetic interference feature matrix into the road condition fingerprint classification model and calculate the similarity value between the real-time road condition features and the various standard road condition fingerprint features in the road condition fingerprint classification model.

[0178] The road surface type recognition module 330 is used to compare the size of the similarity values ​​and select the road surface type corresponding to the standard road condition fingerprint with the largest similarity value as the current road surface type recognition result;

[0179] The adhesion coefficient query module 340 is used to query the pre-stored mapping relationship table between road surface type and adhesion coefficient based on the current road surface type identification result, and output the corresponding road surface adhesion coefficient estimate.

[0180] like Figure 9 As shown, the trajectory stability evaluation module 400 includes:

[0181] The safety index calculation module 410 is used to obtain the vehicle's current speed, steering angle and acceleration driving state parameters, and calculate the maximum safe speed, minimum braking distance and critical steering angle by combining the identified current road surface type and road surface adhesion coefficient estimate.

[0182] The vehicle stability boundary envelope construction module 420 is used to construct the vehicle stability boundary envelope using the maximum safe vehicle speed, minimum braking distance and critical steering angle, and predict the stability boundary of the driving trajectory by comparing the relative position of the current driving state with the vehicle stability boundary envelope in real time.

[0183] The trajectory evaluation module 430 is used to generate a trajectory stability evaluation result containing stability level and adjustment suggestions when it detects that the current driving state is in line with and exceeds the stability boundary of the driving trajectory.

[0184] like Figure 10 As shown, the risk assessment module 500 includes:

[0185] The anomaly detection module 510 is used to monitor the spectral spread width and third harmonic energy change rate of electromagnetic interference noise in real time within a preset frequency band. When the spectral spread width exceeds the reference value of 100kHz and the third harmonic energy change rate exceeds 40dB / ms, it is determined that there are abnormal spectral characteristics.

[0186] The wheel slip risk level calculation module 520 is used to obtain the current wheel speed signal and yaw rate of the vehicle when abnormal spectrum characteristics are determined to exist, and to calculate the wheel slip risk level.

[0187] The warning triggering module 530 is used to trigger a warning signal based on the wheel slippage risk level.

[0188] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0189] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

[0190] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for sensing a driving state of an electric vehicle, characterized by, The method comprises: Continuously monitoring the electromagnetic compatibility background noise of the vehicle electronic control unit, collecting electromagnetic interference signals, and converting them into digitized electromagnetic interference noise spectrum data; Performing time-frequency analysis on the electromagnetic interference noise spectrum data, extracting a set of characteristic parameters, and generating a feature matrix, matching and training the feature matrix with a typical road condition sample library to establish a road condition fingerprint classification model; Inputting the real-time acquired electromagnetic interference feature matrix into the road condition fingerprint classification model, calculating the similarity of the real-time road condition with each type of standard road condition fingerprint in the road condition fingerprint classification model, identifying the current road surface type, and outputting the corresponding road surface adhesion coefficient estimate value; According to the identified current road surface type and road surface adhesion coefficient estimate value, combining the current vehicle driving state parameters, calculating the vehicle safety parameters, predicting the stability boundary of the driving trajectory, and generating the trajectory stability evaluation result; Real-time monitoring of the dynamic change characteristics of electromagnetic interference noise in the preset frequency band, identifying whether there is a spectrum diffusion phenomenon and a specific harmonic energy mutation, if there is, combining the current vehicle driving state parameters to judge the wheel slip risk level, and triggering a graded warning signal; The road condition fingerprint classification model is established, specifically including: For the electromagnetic interference noise spectrum data, extract the power spectral density feature vector as the frequency domain feature, and calculate the peak factor, pulse factor and kurtosis index of the signal as the time domain statistical feature; Combine the power spectral density feature vector, harmonic component distribution and time domain statistical feature to form an initial feature matrix, and perform dimension reduction processing on the initial feature matrix to obtain a reduced feature matrix; Unsupervised learning and training the feature matrix with the standard road condition samples in the typical road condition sample library to establish a road condition fingerprint classification model containing five standard road conditions of asphalt pavement, concrete pavement, gravel pavement, snow-covered pavement and icy pavement; The identification of the current road surface type and the output of the corresponding road surface adhesion coefficient estimate value specifically includes: Real-time acquisition of the current electromagnetic interference signals of the vehicle and conversion into real-time electromagnetic interference feature matrix; Input the real-time electromagnetic interference feature matrix into the road condition fingerprint classification model to calculate the similarity value of the real-time road condition feature and the standard road condition fingerprint feature in the road condition fingerprint classification model; Compare the similarity values to select the road surface type corresponding to the standard road condition fingerprint with the largest similarity value as the current road surface type identification result; According to the current road surface type identification result, query the pre-stored road surface type and adhesion coefficient mapping relationship table, and output the corresponding road surface adhesion coefficient estimate value.

2. The method of claim 1, wherein, The collection of electromagnetic interference signals specifically includes: Synchronously collecting the conducted emission signals and radiation emission signals generated by each electronic control unit of the vehicle within the frequency band of 150 kHz to 1 GHz to obtain the original electromagnetic interference signals; Pretreatment of the original electromagnetic interference signals and signal acquisition, conversion of the pretreated original electromagnetic interference signals into digitized electromagnetic interference noise spectrum data.

3. The method of claim 1, wherein, The generation of the trajectory stability evaluation result specifically includes: Obtain the current vehicle speed, steering angle and acceleration state parameters, combine the recognized current road type and the road adhesion coefficient estimate value, calculate the maximum safe speed, the minimum braking distance and the critical steering angle; Use the maximum safe speed, the minimum braking distance and the critical steering angle to construct the vehicle stability boundary envelope, and by comparing the relative position of the current driving state and the vehicle stability boundary envelope in real time, the stability boundary of the driving trajectory is predicted; When it is detected that the current driving state is attached to or exceeds the stability boundary of the driving trajectory, a trajectory stability evaluation result containing the stability level and adjustment suggestions is generated.

4. The method of claim 3, wherein, The identification of whether there is spectrum spreading phenomenon and specific harmonic energy mutation specifically includes: Real-time monitoring of the spectrum spreading width and the third harmonic energy change rate of the electromagnetic interference noise in the preset frequency band, when the spectrum spreading width exceeds the 100kHz reference value and the third harmonic energy change rate exceeds 40dB / ms, it is determined that there is an abnormal spectrum feature; When it is determined that there is an abnormal spectrum feature, the current wheel speed signal and yaw rate of the vehicle are obtained, and the wheel slip risk level is calculated; According to the wheel slip risk level, a warning signal is triggered.

5. An electric vehicle running state sensing system characterized by comprising: The system includes: A data acquisition module for continuously monitoring the electromagnetic compatibility background noise of the vehicle electronic control unit, collecting electromagnetic interference signals, and converting them into digitized electromagnetic interference noise spectrum data; A time-frequency analysis and feature extraction module for time-frequency analysis of the electromagnetic interference noise spectrum data, extracting a set of feature parameters, and generating a feature matrix, matching the feature matrix with a typical road condition sample library, and establishing a road condition fingerprint classification model; A real-time road condition recognition module for inputting the real-time acquired electromagnetic interference feature matrix into the road condition fingerprint classification model, calculating the similarity of the real-time road condition and each standard road condition fingerprint in the road condition fingerprint classification model, recognizing the current road type, and outputting the corresponding road adhesion coefficient estimate value; A trajectory stability evaluation module for calculating vehicle safety parameters based on the recognized current road type and road adhesion coefficient estimate value, in combination with the current vehicle driving state parameters, predicting the stability boundary of the driving trajectory, and generating a trajectory stability evaluation result; A risk assessment module for real-time monitoring of the dynamic change characteristics of the electromagnetic interference noise in the preset frequency band, identifying whether there is spectrum spreading phenomenon and specific harmonic energy mutation, and if so, combining the current vehicle driving state parameters to judge the wheel slip risk level and trigger a graded warning signal; The execution steps of the time-frequency analysis and feature extraction module include: For the electromagnetic interference noise spectrum data, extract the power spectral density feature vector as the frequency domain feature, and calculate the peak factor, pulse factor and kurtosis index of the signal as the time domain statistical feature; Combine the power spectral density feature vector, harmonic component distribution and time domain statistical feature to form an initial feature matrix, and perform dimension reduction processing on the initial feature matrix to obtain a reduced feature matrix. The feature matrix is unsupervised learning trained with standard road condition samples in a typical road condition sample library, and a road condition fingerprint classification model containing five standard road conditions of asphalt pavement, concrete pavement, gravel pavement, snow-covered pavement and icy pavement is established; The execution steps of the real-time road condition recognition module include: Real-time electromagnetic interference signals of the vehicle are collected and converted into a real-time electromagnetic interference feature matrix; The real-time electromagnetic interference feature matrix is input into the road condition fingerprint classification model, and a similarity value of the real-time road condition feature and each standard road condition fingerprint feature in the road condition fingerprint classification model is calculated; The similarity values are compared, and the road surface type corresponding to the standard road condition fingerprint with the largest similarity value is selected as the current road surface type recognition result; According to the current road surface type recognition result, a pre-stored road surface type and adhesion coefficient mapping relationship table is queried, and a corresponding road surface adhesion coefficient estimation value is output.

6. The system of claim 5, wherein, The trajectory stability evaluation module includes: A safety index calculation module is configured to obtain the current vehicle speed, steering angle and acceleration driving state parameters, combine the recognized current road surface type and the road surface adhesion coefficient estimation value, and calculate the maximum safe speed, the minimum braking distance and the critical steering angle. A vehicle stability boundary envelope line construction module is configured to construct a vehicle stability boundary envelope line using the maximum safe speed, the minimum braking distance and the critical steering angle, and predict the stability boundary of the driving trajectory by comparing the relative positions of the current driving state and the vehicle stability boundary envelope line in real time. A trajectory evaluation module is configured to generate a trajectory stability evaluation result containing a stability level and an adjustment suggestion when it is detected that the current driving state adheres to or exceeds the stability boundary of the driving trajectory.

7. The system of claim 6, wherein, The risk evaluation module includes: An abnormality determination module is configured to monitor the frequency spectrum diffusion width and the third harmonic energy change rate of the electromagnetic interference noise in a preset frequency band in real time, and determine that there is an abnormal spectrum feature when it is detected that the frequency spectrum diffusion width exceeds the 100kHz reference value and the third harmonic energy change rate exceeds 40dB / ms. A wheel slip risk level calculation module is configured to obtain the current wheel speed signal and yaw rate of the vehicle when it is determined that there is an abnormal spectrum feature, and calculate the wheel slip risk level. A warning triggering module is configured to trigger a warning signal according to the wheel slip risk level.

Images