Booster station vehicle detection system

Abstract

The invention relates to the technical field of data processing, in particular to a booster station vehicle detection system which comprises an environment perception optimization module, a joint denoising module, a data fusion module, a threshold value correction module and a closed loop feedback module. The environment perception optimization module generates an anti-interference node deployment map through a genetic algorithm spatial layout model, and optimizes view field overlapping parameters of the laser radar and the ultrasonic sensor; a joint denoising module calls a genetic optimization filter coefficient based on the electromagnetic resonance frequency band mark vector, and executes time-frequency domain joint filtering processing; and the closed-loop feedback module constructs an interference source fingerprint characteristic spectrum and an evolvable alarm rule set, and reversely optimizes sensor layout parameters and decision logic through disposal effect data. And the data circulation control module synchronizes genetic algorithm parameters and state data among multiple modules, so that full-link collaborative optimization is realized, a false alarm phenomenon under a dynamic working condition is effectively inhibited, and the detection precision and the anti-interference capability are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a vehicle detection system for a booster station. Background Art

[0002] The vehicle detection system for a booster station is an important facility to ensure the safe operation of the booster station and plays a key role in the warning scenario of the vehicle height detection device in the booster station. This system mainly consists of a data acquisition module, an analysis and processing module, and an alarm module. The data acquisition module usually uses devices such as lidar and ultrasonic sensors to obtain the height data of vehicles entering the booster station in real time. The analysis and processing module will compare and analyze the collected vehicle height data with a pre-set safety height threshold. When the vehicle height exceeds the safety threshold, the height detection device will trigger an alarm. The alarm module will inform relevant staff in a timely manner by means of sound and light signals, text message notifications, etc. that there is a vehicle with excessive height entering the booster station. The staff can quickly take corresponding measures based on the detailed information such as the vehicle position and height provided by the system, such as guiding the vehicle to leave safely and checking the loading situation of the vehicle, to avoid safety accidents such as colliding with equipment and damaging lines caused by the excessive height of the vehicle, thereby effectively maintaining the normal operation order and equipment safety of the booster station.

[0003] The technical pain point of the vehicle detection system for a booster station in data processing lies in the real-time fusion of multi-source heterogeneous sensor data and the suppression of abnormal interference in a complex environment. In the scenario of dynamic vehicle passage, the height data collected by devices such as lidar and ultrasonic sensors are easily affected by factors such as electromagnetic interference, temperature and humidity changes, and dynamic adjustment of vehicle posture, resulting in noise fluctuations and instantaneous distortions in the original data. Existing algorithms have insufficient recognition ability for non-stationary random interference and are difficult to effectively distinguish the real vehicle contour from instantaneous environmental noise, which may cause false alarms or missed detections and affect the accuracy of alarm judgment. Summary of the Invention

[0004] Aiming at the deficiencies of the prior art, the present invention provides a vehicle detection system for a booster station to solve the problem of high false alarm rate of vehicle height detection caused by insufficient fusion accuracy of multi-source sensor data under complex electromagnetic environments and dynamic working conditions.

[0005] To solve the above technical problems, the specific technical solutions of the present invention are as follows: A vehicle detection system for a booster station provided by the present invention includes an environmental perception optimization module, a joint denoising module, a data fusion module, a threshold correction module, and a closed-loop feedback module, wherein: The environment perception optimization module is deployed in the entrance area of the step-up substation. It is used to obtain the electromagnetic spectrum feature data collected by the distributed electromagnetic field intensity monitoring unit and the field of view overlap rate parameters of the lidar and ultrasonic sensors. It generates sensor topology configuration parameters through the genetic algorithm space layout model and transmits the deployment map including the anti-interference node position information to the joint denoising module. The joint denoising module receives the deployment map, lidar point cloud data, and ultrasonic echo signals, combines the electromagnetic spectrum feature data and inputs it into the adaptive filtering module. It performs time-frequency domain joint filtering processing through the electromagnetic resonance frequency band marked data-driven multi-band filtering coefficient generator, outputs the denoised vehicle height data and electromagnetic interference marking vector, and transmits the height data and marking vector to the data fusion module. The data fusion module, based on the electromagnetic field intensity gradient change rate data stream, calls the Kalman filter gain coefficient in the genetic optimization weight allocation strategy to generate a dynamic weight allocation matrix for motion compensation and coordinate alignment of the denoised lidar point cloud data and ultrasonic data, outputs the fused vehicle three-dimensional contour data and confidence evaluation index, and transmits the fused data stream and confidence index to the threshold correction module; The threshold correction module receives the fused data stream and the electromagnetic field intensity - motion state coupling parameter. It analyzes the mechanical vibration component and electromagnetic pulse interference characteristics in the vehicle acceleration through the genetic algorithm compensation coefficient generation model, combines the historical false alarm event library to generate an anti-interference composite compensation coefficient, and uses a sliding window mechanism to dynamically correct the safety height threshold, outputting the corrected height deviation value and threshold trigger flag bit to the closed-loop feedback module; The closed-loop feedback module receives the corrected height deviation value, equipment operation state data, and electromagnetic interference type recognition result. It generates an interference source spatial position feature map through the interference source fingerprint modeling unit, combines the mutable alarm rule set in the genetic rule evolution engine to generate an alarm decision instruction and disposal suggestion, and reversely feeds back the actual disposal effect data to the fitness evaluation link of the genetic algorithm space layout model in the environment perception optimization module.

[0006] Furthermore, for the vehicle detection system in the step-up substation of the present invention, the environment perception optimization module includes: An electromagnetic spectrum acquisition unit, which is used to obtain the electromagnetic field intensity gradient data and spectrum features in the entrance area of the step-up substation in real time; A chromosome encoding unit, which inputs the lidar field of view overlap rate parameter, ultrasonic sensor installation angle parameter, and the spectrum features into the genetic algorithm space layout model to generate a sensor topology gene encoding sequence; A fitness evaluation unit, which calculates the signal-to-noise ratio improvement rate and sensor data complementarity index of the gene encoding sequence based on the signal attenuation data set generated by the electromagnetic interference simulator; A topology configuration unit iteratively optimizes the gene coding sequence through a genetic algorithm, outputs optimal layout parameters including the positions of anti-interference nodes, and generates a sensor deployment topology map.

[0007] Further, in the vehicle detection system for a booster station according to the present invention, the joint denoising module includes: A feature recognition layer performs a fast Fourier transform on the electromagnetic spectrum feature data and extracts an electromagnetic resonance frequency band marker vector. A gene sequence library stores the frequency band filtering coefficients and temperature and humidity compensation parameters optimized by the genetic algorithm spatial layout model of the environmental perception optimization module. A joint filter calls the matching filtering coefficients from the gene sequence library according to the electromagnetic resonance frequency band marker vector and performs time-frequency domain joint filtering processing on the lidar point cloud data and the ultrasonic echo signal.

[0008] Further, in the vehicle detection system for a booster station according to the present invention, the data fusion module includes: A confidence calculation unit calculates the confidence decay factor of each sensor node based on the electromagnetic field strength gradient change rate data stream. A motion compensation unit calls the Kalman filter gain coefficient in the genetic optimization weight allocation strategy to perform motion trajectory prediction compensation on the dynamic vehicle data. A spatio-temporal calibration engine aligns the coordinates of the denoised lidar point cloud data and ultrasonic data according to the dynamic weight allocation matrix and outputs the fused three-dimensional contour data with timestamp synchronization.

[0009] Further, in the vehicle detection system for a booster station according to the present invention, the threshold correction module includes: A coupling analysis unit analyzes the mechanical vibration component in the vehicle acceleration data and extracts the electromagnetic pulse interference frequency band characteristics. A gene chain coding unit inputs the mechanical vibration component and the interference pattern characteristics in the historical false alarm event library into the genetic algorithm compensation model to generate an anti-interference compensation coefficient gene chain. A trend verification unit uses a sliding window mechanism to perform continuous frame data trend matching verification on the safety height threshold corrected by the anti-interference compensation coefficient.

[0010] Further, in the vehicle detection system for a booster station according to the present invention, the closed-loop feedback module includes: A fingerprint modeling unit receives the threshold trigger flag bit and the device operation status data output by the threshold correction module, performs correlation modeling on the electromagnetic interference type and the spatial position, and generates a feature map including the spatial coordinates of the interference source. A rule evolution engine encodes the interference source coordinate data in the feature map and the disposal records in the historical false alarm event library into a genetic rule evolution model to generate an adaptable and adjustable alarm logic rule set; A feedback adaptation unit reversely transmits the actual disposal effect data of the alarm decision instruction to the fitness evaluation unit of the genetic algorithm space layout model of the environment perception optimization module, for optimizing the iterative parameters of the sensor topology gene coding sequence.

[0011] Furthermore, the vehicle detection system for a booster station according to the present invention further includes a data flow control module, which is used for: Establishing a real-time transmission link for the electromagnetic field strength monitoring data from the electromagnetic spectrum acquisition unit of the environment perception optimization module to the feature recognition layer of the joint denoising module; Configuring a synchronous update channel for the sensor weight parameters optimized by the genetic algorithm space layout model between the spatio-temporal calibration engine of the data fusion module and the gene chain coding unit of the threshold correction module; Controlling the spatio-temporal correlation transmission path of the vehicle motion acceleration data from the output end of the joint denoising module to the motion compensation unit of the data fusion module and the coupling analysis unit of the threshold correction module; Managing the reverse return channel of the disposal effect data generated by the feedback adaptation unit of the closed-loop feedback module to the fitness evaluation unit of the environment perception optimization module, to form a closed-loop optimization link for the genetic algorithm parameters.

[0012] Advantages of the present invention; The present invention effectively reduces the false alarm rate of vehicle height detection in a booster station through a multi-layer collaborative processing mechanism, and improves the detection accuracy in a complex electromagnetic environment. The environment perception optimization module dynamically generates an anti-interference node deployment map based on a genetic algorithm, optimizes the multi-sensor spatial layout and the redundancy of the field of view coverage, and enhances the active suppression ability of the hardware system against electromagnetic interference; the joint denoising module accurately separates the electromagnetic resonance band noise and the real vehicle signal through time-frequency domain joint filtering and genetic optimization of the filter coefficient call, and reduces the instantaneous distortion interference in the original data; the data fusion module dynamically allocates sensor weights in combination with the electromagnetic field strength gradient change rate, and uses motion compensation and spatio-temporal calibration technologies to achieve high-precision spatio-temporal alignment of multi-source heterogeneous data, and improves the confidence of the fused data; the threshold correction module analyzes the coupling characteristics of mechanical vibration and electromagnetic interference, generates an anti-interference compensation coefficient through a genetic algorithm and uses a sliding window mechanism to verify the threshold correction trend, and suppresses the misjudgment risk under dynamic working conditions; the closed-loop feedback module constructs an interference source fingerprint feature map and an evolvable alarm rule set, and reversely optimizes the sensor topology parameters and decision logic through the disposal effect data, forming a continuous iterative improvement of the detection accuracy and anti-interference ability, and finally realizing multi-dimensional collaborative false alarm rate control. Description of the drawings

[0013] To more clearly illustrate the technical solutions of the present invention, the following will briefly introduce the drawings required in the embodiments. Obviously, for those of ordinary skill in the art, other drawings can also be obtained based on the drawings without creative efforts.

[0014] Figure 1 It is a system architecture diagram of a vehicle detection system for a booster station provided by an embodiment of the present invention. Specific embodiments

[0015] To make the objectives, technical solutions, and advantages of the present invention clearer, the following will clearly and completely describe the technical solutions of the present invention in conjunction with specific embodiments of the present invention and the corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention. The following will detail the technical solutions provided by each embodiment of the present invention with reference to the drawings. To better understand the objectives of the present invention, the following will further describe the present invention in detail.

[0016] Please refer to Figure 1 , a vehicle detection system for a booster station provided by the present invention includes an environmental perception optimization module, a joint denoising module, a data fusion module, a threshold correction module, and a closed-loop feedback module, where: The environmental perception optimization module is deployed in the entrance area of the booster station and is used to obtain the electromagnetic spectrum feature data collected by the distributed electromagnetic field strength monitoring unit and the field of view overlap rate parameters of the lidar and ultrasonic sensors. The sensor topology configuration parameters are generated through the genetic algorithm spatial layout model, and the deployment map including the anti-interference node position information is transmitted to the joint denoising module; the joint denoising module receives the deployment map, lidar point cloud data, and ultrasonic echo signals, and inputs them into the adaptive filtering module in combination with the electromagnetic spectrum feature data. The electromagnetic resonance frequency band marked data drives the multi-band filtering coefficient generator to perform time-frequency domain joint filtering processing, and outputs the denoised vehicle height data and electromagnetic interference marking vector, and transmits the height data and marking vector to the data fusion module; the data fusion module, based on the electromagnetic field strength gradient change rate data stream, calls the Kalman filter gain coefficient in the genetic optimization weight allocation strategy to generate a dynamic weight allocation matrix to perform motion compensation and coordinate alignment on the denoised lidar point cloud data and ultrasonic data, and outputs the fused vehicle three-dimensional contour data and confidence evaluation index, and transmits the fused data stream and confidence index to the threshold correction module; The threshold correction module receives the fused data stream and the electromagnetic field strength - motion state coupling parameter, analyzes the mechanical vibration component and the electromagnetic pulse interference characteristics in the vehicle acceleration through a genetic algorithm compensation coefficient generation model, generates an anti-interference composite compensation coefficient in combination with the historical false alarm event library, dynamically corrects the safety height threshold by adopting a sliding window mechanism, and outputs a corrected height deviation value and a threshold trigger flag bit to the closed-loop feedback module; The closed-loop feedback module receives the corrected height deviation value, the device operation state data, and the electromagnetic interference type recognition result, generates an interference source spatial position feature map through an interference source fingerprint modeling unit, generates an alarm decision instruction and a disposal suggestion in combination with the mutable alarm rule set in the genetic rule evolution engine, and reversely feeds back the actual disposal effect data to the fitness evaluation link of the genetic algorithm spatial layout model of the environment perception optimization module.

[0017] A vehicle detection system for a booster station provided by the present invention realizes vehicle height detection and interference suppression in a complex environment through a multi-layer data processing mechanism. The environment perception optimization module is deployed in the entrance area of the booster station, and real-time electromagnetic spectrum feature data is collected through a distributed electromagnetic field strength monitoring unit, and at the same time, the field of view overlap rate parameter of the lidar and the ultrasonic sensor is obtained. The above parameters are input into the genetic algorithm spatial layout model for encoding to generate a gene encoding sequence including the sensor installation angle, the field of view coverage range, and the anti-interference node position. Based on the signal attenuation data set generated by the electromagnetic interference simulator, the signal-to-noise ratio improvement rate and the multi-sensor data complementarity index of each gene sequence are calculated, and the optimal sensor topology configuration parameter is output through iterative evolution, forming an anti-interference node deployment map and transmitting it to the subsequent module.

[0018] The joint denoising module receives the deployment map and the original sensor data output by the environment perception optimization module, and performs interference suppression processing in combination with the real-time electromagnetic spectrum features. The feature recognition layer performs a fast Fourier transform on the electromagnetic spectrum data to extract an electromagnetic resonance marker vector in a specific frequency band. The gene sequence library stores the frequency band filtering coefficients and the temperature and humidity compensation parameters optimized by the genetic algorithm. The joint filter dynamically calls the matching filtering coefficients according to the electromagnetic resonance marker vector, and performs time-frequency domain joint filtering on the lidar point cloud data and the ultrasonic echo signal. The denoised vehicle height data and the electromagnetic interference marker vector are synchronously transmitted to the data fusion module, retaining the spatio-temporal distribution information of the interference characteristics.

[0019] The data fusion module dynamically adjusts the fusion weights of multi-source sensor data based on the data stream of the electromagnetic field strength gradient change rate. The confidence calculation unit calculates the confidence attenuation factor of each sensor node according to the field strength gradient change rate, and generates a dynamic weight distribution matrix in combination with the Kalman filter gain coefficient in the genetic optimization weight distribution strategy. The motion compensation unit uses the Kalman filter algorithm to predict the motion trajectory of the dynamic vehicle data. The spatio-temporal calibration engine aligns the coordinates of the denoised lidar and ultrasonic data according to the weight matrix, and outputs the three-dimensional contour data and confidence evaluation index with timestamp synchronization, providing a fusion data basis for threshold correction.

[0020] The threshold correction module analyzes the coupling characteristics of the vehicle motion state and electromagnetic interference. The coupling analysis unit separates the mechanical vibration component from the vehicle acceleration data, and at the same time extracts the frequency band characteristics of the electromagnetic pulse interference. The gene chain encoding unit associates and encodes the above characteristics with the interference patterns in the historical false alarm event library, and generates a gene chain of anti-interference compensation coefficients through the genetic algorithm compensation model. The sliding window mechanism performs continuous frame data trend matching verification on the compensated safety height threshold, and outputs the corrected height deviation value and the threshold trigger flag bit, suppressing the false alarm phenomenon caused by transient interference.

[0021] The closed-loop feedback module receives the threshold correction result and the device operation state data, and constructs a feedback link for interference source location and alarm decision-making. The fingerprint modeling unit performs associated modeling on the spatial position of the interference source according to the threshold trigger flag bit and the electromagnetic interference type recognition result, and generates a feature map containing coordinate information. The rule evolution engine encodes the interference source data in the feature map and the historical disposal records into the genetic rule evolution model together, and generates an alarm logic rule set that can be adaptively adjusted. The feedback adaptation unit reversely transmits the actual disposal effect data to the fitness evaluation unit of the environmental perception optimization module, driving the iterative optimization of the sensor topology gene coding sequence, and forming a closed-loop control mechanism for electromagnetic interference suppression and detection accuracy improvement.

[0022] The above-mentioned modules achieve cross-layer collaboration through the data flow control module. The electromagnetic field strength monitoring data is transmitted from the environmental perception optimization module to the joint denoising module along the feedforward link. The sensor weight parameters optimized by the genetic algorithm are synchronously updated between the data fusion and threshold correction modules. The vehicle motion state data runs through the denoising, fusion, and threshold correction processing links along the spatio-temporal association path. The disposal effect data is injected into the parameter optimization link through the reverse return channel. The multi-dimensional data interaction and genetic algorithm parameter transfer mechanism ensure the high-precision fusion of multi-source sensor data and dynamic interference suppression in a complex electromagnetic environment.

[0023] Specifically, for the booster station vehicle detection system described in the present invention, the environmental perception optimization module includes: An electromagnetic spectrum acquisition unit for obtaining the electromagnetic field strength gradient data and spectrum characteristics in the entrance area of the step-up substation in real time; A chromosome encoding unit that inputs the field of view overlap rate parameter of the lidar, the installation angle parameter of the ultrasonic sensor, and the spectrum characteristics into the genetic algorithm spatial layout model to generate a sensor topology gene encoding sequence; A fitness evaluation unit that calculates the signal-to-noise ratio improvement rate and the sensor data complementarity index of the gene encoding sequence based on the signal attenuation data set generated by the electromagnetic interference simulator; A topology configuration unit that iteratively optimizes the gene encoding sequence through the genetic algorithm, outputs the optimal layout parameters including the positions of anti-interference nodes, and generates a sensor deployment topology map.

[0024] The environmental perception optimization module of the present invention realizes the dynamic optimization of sensor layout through multi-level data processing. The electromagnetic spectrum acquisition unit distributes electromagnetic field strength monitoring nodes in the entrance area of the step-up substation to collect electromagnetic field strength gradient data and spectrum characteristics in real time. The monitoring data includes the electromagnetic interference intensity distribution and frequency domain energy characteristics at different spatial positions. The collected electromagnetic field strength gradient data, the field of view overlap rate parameter of the lidar, and the installation angle parameter of the ultrasonic sensor are jointly input into the chromosome encoding unit. Among them, the field of view overlap rate parameter characterizes the detection area coverage redundancy of multiple lidars, and the installation angle parameter defines the elevation angle and azimuth angle information of the ultrasonic sensor. The above parameters are gene-encoded through the genetic algorithm spatial layout model to generate a gene encoding sequence including sensor spatial coordinates, detection directions, and anti-interference priority parameters.

[0025] The fitness evaluation unit constructs a signal attenuation data set based on the electromagnetic interference simulator to simulate the electromagnetic interference effects of different sensor layout schemes under typical working conditions. The signal-to-noise ratio improvement rate index is obtained by comparing the effective signal strength and the noise floor strength of the layout scheme corresponding to the gene encoding sequence in the interference environment. The sensor data complementarity index is generated by analyzing the synergistic effects of the lidar and the ultrasonic sensor in the dimensions of spatial coverage, sampling frequency, and anti-interference ability. The evaluation results are fed back to the genetic algorithm spatial layout model to drive the chromosome encoding unit to perform crossover and mutation operations on the gene encoding sequence.

[0026] The topology configuration unit executes the iterative optimization process of the genetic algorithm, using the signal-to-noise ratio improvement rate and the data complementarity index output by the fitness evaluation unit as the evolutionary direction constraint conditions. In each iteration process, the sensor coordinate parameters and the anti-interference node position parameters in the gene encoding sequence are adaptively adjusted. The optimized layout scheme synchronously generates a topology map including sensor installation coordinates, detection direction calibration data, and anti-interference node deployment suggestions. The finally output optimal layout parameters integrate the multi-sensor spatial distribution optimization results and the anti-interference node configuration strategy, providing a hardware deployment basis for anti-electromagnetic interference for subsequent modules.

[0027] Specifically, for the vehicle detection system of the booster station described in the present invention, the joint denoising module includes: A feature recognition layer that performs a fast Fourier transform on the electromagnetic spectrum feature data to extract an electromagnetic resonance frequency band marker vector; A gene sequence library that stores the frequency band filtering coefficients and temperature-humidity compensation parameters optimized by the genetic algorithm spatial layout model of the environment perception optimization module; A joint filter that calls the matching filtering coefficients from the gene sequence library according to the electromagnetic resonance frequency band marker vector and performs time-frequency domain joint filtering processing on the lidar point cloud data and the ultrasonic echo signal.

[0028] The joint denoising module realizes the denoising of sensor data in a complex electromagnetic environment through multi-stage processing. The feature recognition layer performs a fast Fourier transform on the electromagnetic spectrum feature data, converts the time-domain signal into a frequency-domain energy distribution map, and identifies the electromagnetic resonance characteristics of a specific frequency band. Peak detection is performed on the spectrum data by setting a dynamic energy threshold, and the center frequency, bandwidth, and energy intensity parameters of the resonance frequency band of electromagnetic interference are extracted to generate a feature vector containing frequency band marker information. This feature vector characterizes the frequency-domain distribution characteristics of the current electromagnetic interference and provides a frequency band selection basis for subsequent filtering processing.

[0029] The gene sequence library stores the frequency band filtering coefficients and temperature-humidity compensation parameters optimized by the genetic algorithm spatial layout model. The frequency band filtering coefficients are dynamically adjusted according to the anti-interference node deployment map generated by the environment perception optimization module and include the center frequency, attenuation slope, and stopband width parameters of the band-stop filter. The temperature-humidity compensation parameters are generated through training with historical environmental data to establish an association mapping relationship between temperature-humidity changes and sensor signal attenuation. The gene sequence library adopts a hierarchical index structure, uses the frequency band characteristics in the electromagnetic resonance frequency band marker vector as the retrieval key values, and realizes the fast matching and calling of filtering parameters.

[0030] Based on the matching result of the electromagnetic resonance frequency band marker vector and the gene sequence library, the joint filter constructs a time-frequency domain joint filtering processing architecture. For the lidar point cloud data, a time-domain sliding window mechanism is adopted, combined with the Kalman filtering algorithm to suppress transient pulse interference; for the ultrasonic echo signal, frequency-domain adaptive filtering is applied, and the stopband range is dynamically adjusted according to the matching frequency band filtering coefficients. After the time-frequency domain processing results are weighted and fused, the denoised vehicle height data and the electromagnetic interference marker vector are output. The marker vector retains the frequency band characteristics and spatio-temporal distribution information of the filtered interference, providing a basis for interference traceability for the subsequent data fusion module.

[0031] Specifically, for the vehicle detection system of the booster station described in the present invention, the data fusion module includes: A confidence calculation unit calculates a confidence decay factor for each sensor node based on the electromagnetic field strength gradient change rate data stream. A motion compensation unit calls the Kalman filter gain coefficient in the genetic optimization weight allocation strategy to perform motion trajectory prediction compensation on dynamic vehicle data. A spatio-temporal calibration engine aligns the coordinates of the denoised lidar point cloud data and ultrasonic data according to the dynamic weight allocation matrix, and outputs fused three-dimensional contour data with timestamp synchronization.

[0032] The data fusion module achieves high-precision fusion of sensor data through multi-dimensional data processing. The confidence calculation unit receives the electromagnetic field strength gradient change rate data stream in real time and analyzes the electromagnetic interference intensity fluctuation characteristics at the positions of each sensor node. Based on the negative correlation between the electromagnetic field strength gradient change rate and the stability of sensor data, a calculation model for the confidence decay factor is established to dynamically evaluate the data reliability of lidar and ultrasonic sensors in a time-varying interference environment, and a sequence of confidence decay factors for each node is output.

[0033] The motion compensation unit calls the pre-trained Kalman filter gain coefficient in the genetic optimization weight allocation strategy to perform motion state compensation on dynamic vehicle data. According to the real-time speed and acceleration parameters of the vehicle, a motion trajectory prediction model is constructed, and the time delay deviation of the lidar point cloud data is corrected through the Kalman filter algorithm. At the same time, the detection angle offset caused by the vehicle attitude change of the ultrasonic sensor is compensated to generate motion compensation data aligned with the time axis.

[0034] The spatio-temporal calibration engine receives the denoised multi-source sensor data and the dynamic weight allocation matrix, and performs spatio-temporal dimension data integration. The dynamic weight allocation matrix fuses the confidence decay factor and the Kalman filter gain coefficient, and calculates the data weight ratios of the lidar and ultrasonic sensors on each coordinate axis in three-dimensional space. The coordinate alignment process uses the Euler angle transformation algorithm to unify the multi-sensor coordinate systems, and the timestamp synchronization mechanism is based on the time axis calibration result of the motion compensation data, and outputs the fused vehicle three-dimensional contour data, whose spatial resolution and time consistency meet the input requirements of the threshold correction module.

[0035] Specifically, for the booster station vehicle detection system of the present invention, the threshold correction module includes: A coupling analysis unit analyzes the mechanical vibration components in the vehicle acceleration data and extracts the electromagnetic pulse interference frequency band characteristics. A gene chain encoding unit inputs the mechanical vibration components and the interference pattern characteristics in the historical false alarm event library into the genetic algorithm compensation model to generate an anti-interference compensation coefficient gene chain. A trend verification unit uses a sliding window mechanism to perform continuous frame data trend matching verification on the safety height threshold corrected by the anti-interference compensation coefficient.

[0036] The threshold correction module realizes the dynamic correction of the safety height threshold through multi-level data processing. The coupling analysis unit receives the vehicle acceleration sensor data, uses the wavelet packet decomposition algorithm to separate the mechanical vibration components, and identifies the vibration frequency band characteristics matching the resonance frequency of the vehicle chassis. It synchronously analyzes the electromagnetic pulse interference signal, calculates the interference intensity of a specific frequency band through spectral energy integration, extracts the electromagnetic interference feature vector including the center frequency, pulse width, and repetition period, and establishes a spatio-temporal correlation model of mechanical vibration and electromagnetic interference.

[0037] The gene chain encoding unit performs feature mapping on the vibration frequency band characteristics output by the coupling analysis unit and the interference patterns in the historical false alarm event library. The historical false alarm event library stores the false alarm event data caused by mechanical vibration or electromagnetic interference under different working conditions, including interference waveforms, environmental parameters, and handling records. Through the genetic algorithm compensation model, gene encoding is performed on the vibration characteristics and interference patterns to generate an anti-interference compensation coefficient gene chain including frequency band suppression weights, time decay factors, and spatial compensation intensities, providing a parameterized compensation strategy for threshold correction.

[0038] The trend verification unit dynamically verifies the compensated safety height threshold using a sliding window mechanism. Based on time series analysis, a trend matching model of continuous frame data within the window is constructed, and the deviation index between the corrected threshold and the actual vehicle height data is calculated. When the deviation index within the window exceeds the preset tolerance, the compensation coefficient iteration update of the gene chain encoding unit is triggered, and the interference pattern feature data in the historical false alarm event library is updated synchronously. The threshold data that passes the verification is output to the closed-loop feedback module to form a closed-loop optimization link for anti-interference parameters and decision-making logic.

[0039] Specifically, for the vehicle detection system in the booster station described in the present invention, the closed-loop feedback module includes: A fingerprint modeling unit that receives the threshold trigger flag bit and device operation status data output by the threshold correction module, performs correlation modeling on the electromagnetic interference type and spatial position, and generates a feature map including the spatial coordinates of the interference source; A rule evolution engine that jointly encodes the interference source coordinate data in the feature map and the handling records in the historical false alarm event library into a genetic rule evolution model to generate an adaptable alarm logic rule set; A feedback adaptation unit that reversely transmits the actual handling effect data of the alarm decision instruction to the fitness evaluation unit of the genetic algorithm spatial layout model in the environment perception optimization module for optimizing the iteration parameters of the sensor topology gene encoding sequence.

[0040] The closed-loop feedback module realizes the adaptive optimization of alarm decision-making through multi-source data interaction. The fingerprint modeling unit receives the threshold trigger flag bit and device operation status data output by the threshold correction module, and analyzes the timestamp, interference intensity, and sensor node position information of the electromagnetic interference event. The spatial clustering algorithm is used to correlate and model the electromagnetic interference type and spatial coordinates, generating an interference source feature map containing the longitude and latitude coordinates of the interference source, the heat map of the interference intensity distribution, and the frequency band feature encoding, providing spatial dimension data support for interference tracing.

[0041] The rule evolution engine calls the disposal record data stored in the historical false alarm event library, including the historical alarm trigger time, the execution effect of the interference suppression measure, and the false alarm determination result. Feature mapping is performed on the spatial coordinate data in the interference source feature map and the disposal record, and the genetic rule evolution model is used to perform gene encoding on the alarm logic rules, generating an evolvable rule set containing the alarm level division rules, the priority of the interference suppression strategy, and the disposal time limit constraint parameters. The evolution direction of the rule set is driven by the statistical distribution of successful suppression cases and false alarm cases in the disposal record, realizing the dynamic adaptation of the alarm logic.

[0042] The feedback adaptation unit collects the actual disposal effect data after the execution of the alarm decision instruction, including the alarm response duration, the time when the interference suppression measure takes effect, and the re-inspection data of the vehicle height detection result. The disposal effect data is converted into the fitness evaluation parameters of the genetic algorithm spatial layout model and reversely transmitted to the fitness evaluation unit of the environment perception optimization module. The fitness evaluation unit dynamically adjusts the crossover probability and mutation step size parameters of the sensor topology gene encoding sequence according to the disposal effect data, optimizing the balance of the sensor layout scheme in the dimensions of anti-electromagnetic interference and data acquisition accuracy, and forming a closed-loop parameter update link from decision execution to layout optimization.

[0043] Specifically, the vehicle detection system of the booster station described in the present invention further includes a data flow control module for: Establishing a real-time transmission link for the electromagnetic field strength monitoring data from the electromagnetic spectrum acquisition unit of the environment perception optimization module to the feature recognition layer of the joint denoising module; Configuring a synchronous update channel for the sensor weight parameters optimized by the genetic algorithm spatial layout model between the spatio-temporal calibration engine of the data fusion module and the gene chain encoding unit of the threshold correction module; Controlling the spatio-temporal correlation transmission path of the vehicle motion acceleration data from the output end of the joint denoising module to the motion compensation unit of the data fusion module and the coupling analysis unit of the threshold correction module; Managing the reverse return channel of the disposal effect data generated by the feedback adaptation unit of the closed-loop feedback module to the fitness evaluation unit of the environment perception optimization module, forming a closed-loop optimization link for the genetic algorithm parameters.

[0044] The data flow control module realizes system-level data collaborative management through a multi-dimensional data transmission channel. A real-time transmission link is established for the electromagnetic field strength monitoring data from the electromagnetic spectrum acquisition unit of the environmental perception optimization module to the feature recognition layer of the joint denoising module. The time-sensitive network protocol is adopted to ensure the low-latency transmission of the electromagnetic spectrum feature data. A data verification mechanism is integrated into the transmission link to perform integrity verification on the electromagnetic field strength gradient data and spectrum features, avoiding the invalidation of subsequent filtering processing caused by packet loss during transmission.

[0045] Configure the synchronous update channel of the sensor weight parameters optimized by the genetic algorithm spatial layout model between the data fusion module and the threshold correction module. The sensor weight parameters include the spatial coverage weights of lidar and ultrasonic sensors, the anti-interference priority coefficients, and the data confidence indicators, and the parameters are synchronously updated in real time between the spatio-temporal calibration engine and the gene chain encoding unit through the publish-subscribe mode. The synchronous update channel adopts a version control mechanism to ensure the dynamic matching of the weight parameters with the current electromagnetic environment state and the sensor layout topology.

[0046] Control the spatio-temporal correlation transmission path of the vehicle motion acceleration data from the output end of the joint denoising module to the motion compensation unit of the data fusion module and the coupling analysis unit of the threshold correction module. A timestamp alignment engine is embedded in the transmission path to perform frame synchronization processing on the denoised vehicle height data and the original acceleration data. The motion compensation unit constructs a vehicle motion trajectory prediction model based on the synchronous data, and the coupling analysis unit extracts the mechanical vibration components in the acceleration data. The two work together to jointly suppress motion interference and electromagnetic interference.

[0047] Manage the reverse return channel of the disposal effect data generated by the feedback adaptation unit of the closed-loop feedback module to the fitness evaluation unit of the environmental perception optimization module. The disposal effect data includes warning response efficiency, interference suppression success rate, and false alarm rate statistical indicators, and is converted into input parameters of the genetic algorithm fitness function through data standardization processing. The fitness evaluation unit dynamically adjusts the crossover and mutation strategies of the sensor topology gene coding sequence according to the returned data, driving the continuous optimization of the anti-interference node deployment map and the sensor layout parameters, and forming a complete data link for the closed-loop update of the genetic algorithm parameters.

[0048] Explanation of the technical features of the present invention: Genetic algorithm spatial layout model: An optimization algorithm that simulates the principle of biological evolution, encodes the sensor layout parameters (such as lidar field of view overlap rate, ultrasonic installation angle) and electromagnetic field strength data into gene sequences, iteratively calculates the signal-to-noise ratio improvement rate and data complementarity index, and outputs the optimal sensor spatial distribution scheme with the best anti-interference ability. This model solves the global optimization problem of sensor layout in complex electromagnetic environments.

[0049] Electromagnetic resonance frequency band marking vector: The electromagnetic interference frequency domain feature data extracted based on the fast Fourier transform, including the center frequency, bandwidth, and energy intensity parameters of the interference frequency band, is used to drive the dynamic call of multi-band filtering coefficients to achieve precise suppression of noise frequency bands.

[0050] Dynamic weight allocation matrix: A weight parameter matrix generated according to the real-time electromagnetic field strength gradient change rate and the sensor confidence decay factor, combined with the genetic optimization weight allocation strategy, is used to quantify the data fusion weights of lidar and ultrasonic sensors on each coordinate axis in three-dimensional space to improve the spatio-temporal alignment accuracy of multi-source data.

[0051] Anti-interference compensation coefficient gene chain: A parameter sequence generated by genetic encoding of mechanical vibration components, electromagnetic pulse interference characteristics, and historical false alarm event data through genetic algorithms, including compensation parameters such as frequency band suppression weights and time decay factors, is used to dynamically correct the safety height threshold to suppress misjudgments.

[0052] Interference source fingerprint feature map: Electromagnetic interference source localization data constructed based on spatial clustering algorithms, integrating interference types, spatial coordinates, and intensity distribution heat map features, provides a spatial dimension basis for alarm rule evolution and interference traceability.

[0053] Data flow control module: The core management unit for realizing cross-module data collaboration, including: Real-time transmission link: Transmits electromagnetic field strength data using the time-sensitive network protocol to ensure low latency and integrity; Synchronous update channel: Realizes the dynamic synchronization of genetic algorithm parameters among multiple modules through the publish-subscribe mode; Spatio-temporal correlation transmission path: Embeds a timestamp alignment engine to ensure the temporal consistency of motion state data in multiple processing links; Reverse return channel: Standardizes the disposal effect data and injects it into the fitness evaluation of the genetic algorithm to form a closed-loop optimization link.

[0054] Technical feature collaboration logic: The anti-interference deployment map output by the environmental perception optimization module drives the call of frequency band filtering parameters of the joint denoising module; the dynamic weight allocation matrix of the data fusion module provides high-precision fusion data for threshold correction; the closed-loop feedback module generates adaptive alarm decisions through the interference source feature map and the rule evolution engine, and reversely optimizes the sensor layout parameters; the data flow control module ensures the collaboration of multi-source data in the transmission, processing, and feedback links. Each technical feature forms a closed loop through genetic algorithm parameter transfer and data interaction, and finally realizes multi-dimensional suppression of the false alarm rate in the dynamic electromagnetic environment.

[0055] Genetic Algorithm Spatial Layout Model: This model is used to optimize the spatial deployment strategy for sensors at the entrance of a substation. By simulating biological evolution, the lidar field of view overlap ratio, ultrasonic sensor installation angle parameters, and real-time electromagnetic spectrum signature data are encoded into genetic sequences. During the iteration process, the model uses signal-to-noise ratio improvement and sensor data complementarity as fitness evaluation metrics to identify the sensor topology configuration with the best anti-interference capabilities. The model then outputs a deployment map that includes anti-interference node locations and field of view coverage. This model addresses the limited adaptability of traditional fixed layouts in dynamic electromagnetic environments.

[0056] Genetic rule evolution model: Integrated into the closed-loop feedback module, it dynamically optimizes the alarm decision logic. This model encodes interference source spatial coordinate data and historical false alarm event handling records into a mutable rule gene chain. Through crossover and mutation operations, it generates an adaptive alarm rule set. The rule set includes alarm level classification, interference suppression priority, and response time parameters. Its evolutionary direction is driven by actual handling effect data, reducing the risk of rule rigidity caused by environmental changes.

[0057] Genetic Algorithm Compensation Model: Applied to the threshold correction module, it generates anti-interference compensation coefficients. This model genetically encodes the mechanical vibration components in vehicle acceleration data, the frequency band characteristics of electromagnetic pulse interference, and interference pattern characteristics from a historical false alarm event database. It outputs a compensation coefficient gene chain consisting of frequency band suppression weights and time decay factors. This model verifies the compensation effect using a sliding window mechanism and dynamically adjusts the threshold correction parameters to mitigate misjudgments caused by the coupling of mechanical vibration and electromagnetic interference.

[0058] A joint time-frequency domain filtering model, implemented in the joint denoising module, suppresses multi-source noise interference. Based on electromagnetic resonance frequency band label vectors, the model dynamically calls frequency band filter coefficients optimized by a genetic algorithm from a gene sequence library. LiDAR point cloud data uses a time-domain sliding window combined with a Kalman filter to suppress transient pulse interference. Ultrasonic echo signals undergo dynamic stopband adjustment through frequency-domain adaptive filtering. This model preserves the spatiotemporal distribution of the filtered interference, providing a denoising foundation for subsequent data fusion.

[0059] Dynamic Weight Allocation Model: Located in the data fusion module, this model quantifies the fusion weights of multi-sensor data. This model calculates the confidence attenuation factor for each sensor node based on the rate of change of the electromagnetic field intensity gradient. Combined with the Kalman filter gain coefficients from the genetic optimization strategy, this model generates a dynamic weight allocation matrix for the three-dimensional spatial coordinate axes. Through motion compensation and spatiotemporal alignment, the coordinate systems and timestamps of the lidar and ultrasonic sensors are unified, outputting high-confidence fused 3D contour data.

[0060] Data Collaboration Management Model: Implemented by the data flow control module to ensure the real-time performance and consistency of cross-module data interaction. The model includes: Real-time Transmission Link: Transmits electromagnetic field strength data using the time-sensitive network protocol and integrates a data verification mechanism to prevent packet loss; Parameter Synchronization Channel: Dynamically synchronizes the sensor weight parameters optimized by the genetic algorithm based on the publish-subscribe mode; Spatio-temporal Association Path: Embeds a timestamp alignment engine to ensure the temporal consistency of vehicle motion state data in multi-module processing; Reverse Flow Channel: Standardizes the disposal effect data and injects it reversely into the fitness evaluation of the genetic algorithm to form a closed-loop optimization link.

[0061] Model Collaboration Logic: The deployment map generated by the environmental perception optimization module drives the filtering parameter call of the joint denoising module; the dynamic weight allocation of the data fusion module provides high-precision input for threshold correction; the closed-loop feedback module improves the system's adaptive ability through rule evolution and parameter reverse optimization; the data collaboration management model ensures the data interaction efficiency of the entire link. Each model solves the core problem of insufficient accuracy in multi-source data fusion under complex electromagnetic environments through genetic algorithm parameter transfer and data flow collaboration, significantly reducing the false alarm rate of vehicle height detection.

[0062] The specific implementation of the present invention is based on the actual application scenario of the entrance area of the booster station, combines complex electromagnetic interference and vehicle dynamic passing requirements, and constructs a vehicle height detection system with multi-module collaboration. In the environmental perception optimization module, the distributed electromagnetic field strength monitoring unit collects electromagnetic spectrum feature data in real time, and the lidar and ultrasonic sensors synchronously obtain the field of view overlap rate and installation angle parameters, which are input into the genetic algorithm spatial layout model for gene coding. The model generates the optimal layout parameters including anti-interference node coordinates and field of view coverage range by iteratively optimizing the sensor topology gene coding sequence, and outputs the sensor deployment topology map to the joint denoising module to provide a basis for anti-interference hardware configuration for subsequent processing.

[0063] After the combined denoising module receives the deployment map and the original sensor data, the feature recognition layer performs a fast Fourier transform on the electromagnetic spectrum data, extracts the electromagnetic resonance frequency band marker vector, and drives the multi-band filter coefficient generator to call the genetically optimized filter parameters stored in the gene sequence library. The lidar point cloud data suppresses transient pulse interference through the time-domain sliding window mechanism, and the ultrasonic echo signal dynamically adjusts the stopband range based on frequency-domain adaptive filtering, outputting the denoised vehicle height data and the electromagnetic interference marker vector, and retaining the spatio-temporal distribution information of the interference characteristics. The data fusion module calculates the confidence decay factor of each sensor node based on the electromagnetic field strength gradient change rate data stream, combines the Kalman filter gain coefficient in the genetically optimized weight allocation strategy to generate a dynamic weight allocation matrix, performs motion compensation and coordinate alignment on the denoised lidar and ultrasonic data, and outputs the time-stamp synchronized fused three-dimensional contour data and the confidence evaluation index, providing high-precision data input for threshold correction.

[0064] The threshold correction module analyzes the vehicle acceleration signal and the electromagnetic field strength - motion state coupling parameter in the fused data stream, separates the mechanical vibration component through wavelet packet decomposition, extracts the electromagnetic pulse interference frequency band characteristics, and inputs them into the genetic algorithm compensation model to generate the anti-interference compensation coefficient gene chain. The sliding window mechanism performs continuous frame data trend matching verification on the compensated safety height threshold, and outputs the corrected height deviation value and the threshold trigger flag bit to the closed-loop feedback module. The closed-loop feedback module constructs the interference source fingerprint feature map according to the threshold trigger flag bit and the device operation state data, combines the handling records in the historical false alarm event library, and generates an adaptable warning logic rule set through the genetic rule evolution engine. The feedback adaptation unit reversely transmits the handling effect data such as the warning response efficiency and the interference suppression success rate to the fitness evaluation unit of the environmental perception optimization module, driving the iterative optimization of the sensor topology gene coding sequence, and forming a continuous improvement in the detection accuracy and anti-interference ability.

[0065] The data transfer control module realizes the whole-system data collaboration through the real-time transmission link, the synchronous update channel, and the reverse return channel. The electromagnetic field strength monitoring data is transmitted to the combined denoising module using the time-sensitive network protocol to ensure low latency and data integrity; the sensor weight parameters optimized by the genetic algorithm are dynamically synchronized between the data fusion and threshold correction modules through the publish-subscribe mode; the vehicle motion acceleration data realizes the timing consistency processing between multiple modules through the spatio-temporal correlation transmission path. The above implementation method effectively suppresses the false alarm phenomenon caused by electromagnetic interference and mechanical vibration through genetic algorithm parameter transfer, dynamic data processing, and closed-loop feedback mechanism, meets the vehicle height detection requirements under the complex working conditions of the booster station, and supports the integrity and feasibility of the technical solutions described in the claims.

[0066] The present invention solves the false alarm rate problem caused by insufficient accuracy of multi-source sensor data fusion in a complex electromagnetic environment through a multi-layer data processing architecture and a dynamic parameter optimization mechanism. The environmental perception optimization module generates an anti-interference node deployment map and sensor topology configuration parameters based on a genetic algorithm spatial layout model. By collecting electromagnetic field strength gradient data and sensor field of view parameters in real time, constructing a gene coding sequence and iteratively optimizing it, the spatial layout of lidar and ultrasonic sensors is dynamically adjusted, improving the data complementarity and anti-interference ability of multi-sensors in an electromagnetic interference environment, and reducing the impact of interference signals on the original data from the hardware deployment level.

[0067] The joint denoising module and the data fusion module cooperate to process multi-source heterogeneous data. The joint denoising module adopts a time-frequency domain joint filtering technology, combines the frequency band filtering coefficients optimized by a genetic algorithm, and suppresses the interference of electromagnetic resonance frequency band noise on lidar point clouds and ultrasonic echo signals. The data fusion module dynamically assigns sensor weights based on the change rate of the electromagnetic field strength gradient, and achieves high-precision alignment of multi-sensor data through motion compensation and spatio-temporal calibration. The combination of the confidence evaluation index and the dynamic weight assignment strategy effectively distinguishes the real vehicle contour from environmental noise, improving the reliability and stability of data fusion.

[0068] The threshold correction module and the closed-loop feedback module form an adaptive optimization link. The threshold correction module analyzes the coupling characteristics of vehicle motion state and electromagnetic interference, generates an anti-interference coefficient gene chain using a genetic algorithm compensation model, and dynamically corrects the safety height threshold in combination with a sliding window mechanism. The closed-loop feedback module injects the disposal effect data reversely into the sensor topology optimization process through interference source fingerprint modeling and a genetic rule evolution engine, driving the continuous iteration of the alarm logic rule set and anti-interference parameters. The data flow control module synchronizes the weight parameters and state data among multiple modules, ensuring the full-link cooperation of the system under dynamic working conditions, and finally achieving multi-dimensional suppression of the false alarm rate.

Claims

1. A vehicle detection system for a booster station, characterized in that, Including: The environmental perception optimization module is used to obtain the electromagnetic spectrum feature data collected by the distributed electromagnetic field strength monitoring unit, as well as the field of view overlap rate parameters of the lidar and ultrasonic sensors, generate sensor topology configuration parameters through the genetic algorithm space layout model, and transmit the deployment map including the anti-interference node position information to the joint denoising module; the joint denoising module performs time-frequency domain joint filtering processing through the electromagnetic resonance frequency band marked data-driven multi-band filtering coefficient generator, outputs the denoised vehicle height data and the electromagnetic interference marked vector, and transmits the height data and the marked vector to the data fusion module; Based on the electromagnetic field strength gradient change rate data stream, the data fusion module calls the Kalman filter gain coefficient in the genetic optimization weight allocation strategy, generates a dynamic weight allocation matrix to perform motion compensation and coordinate alignment on the denoised lidar point cloud data and ultrasonic data, outputs the fused vehicle three-dimensional contour data and the confidence evaluation index, and transmits the fused data stream and the confidence index to the threshold correction module; The threshold correction module receives the fused data stream and the electromagnetic field strength - motion state coupling parameter, analyzes the mechanical vibration component and the electromagnetic pulse interference characteristics in the vehicle acceleration through the genetic algorithm compensation coefficient generation model, combines the historical false alarm event library to generate an anti-interference composite compensation coefficient, dynamically corrects the safety height threshold by using the sliding window mechanism, and outputs the corrected height deviation value and the threshold trigger flag bit to the closed-loop feedback module; The closed-loop feedback module receives the corrected height deviation value, the device operation state data and the electromagnetic interference type recognition result, generates an interference source spatial position feature map through the interference source fingerprint modeling unit, combines the mutable alarm rule set in the genetic rule evolution engine to generate an alarm decision instruction and a disposal suggestion, and reversely feeds back the actual disposal effect data to the fitness evaluation link of the genetic algorithm space layout model of the environmental perception optimization module.

2. The vehicle detection system for a step-up substation according to claim 1, wherein The environmental perception optimization module includes: The electromagnetic spectrum acquisition unit is used to obtain the electromagnetic field strength gradient data and spectrum characteristics of the entrance area of the step-up substation in real time; The chromosome encoding unit inputs the lidar field of view overlap rate parameter, the ultrasonic sensor installation angle parameter and the spectrum characteristics into the genetic algorithm space layout model to generate a sensor topology gene encoding sequence; The fitness evaluation unit calculates the signal-to-noise ratio improvement rate and the sensor data complementarity index of the gene encoding sequence based on the signal attenuation data set generated by the electromagnetic interference simulator; The topology configuration unit iteratively optimizes the gene encoding sequence through the genetic algorithm, outputs the optimal layout parameters including the anti-interference node position, and generates a sensor deployment topology map.

3. The step-up substation vehicle detection system according to claim 1, characterized in that The joint denoising module includes: The feature recognition layer performs a fast Fourier transform on the electromagnetic spectrum feature data to extract the electromagnetic resonance frequency band marked vector; The gene sequence library stores the band filtering coefficients and temperature and humidity compensation parameters optimized by the genetic algorithm space layout model of the environmental perception optimization module; A combined filter that calls matching filter coefficients from a gene sequence library according to the electromagnetic resonance frequency band marking vector and performs time-frequency domain combined filtering processing on lidar point cloud data and ultrasonic echo signals.

4. The vehicle detection system for a booster station according to claim 1, characterized in that The data fusion module includes: A confidence calculation unit that calculates the confidence decay factor of each sensor node based on the electromagnetic field strength gradient change rate data stream; A motion compensation unit that calls the Kalman filter gain coefficient in the genetic optimization weight allocation strategy to perform motion trajectory prediction compensation on dynamic vehicle data; A spatio-temporal calibration engine that aligns the coordinates of the denoised lidar point cloud data and ultrasonic data according to the dynamic weight allocation matrix and outputs fused three-dimensional contour data with synchronized timestamps.

5. The vehicle detection system for a step-up substation according to claim 1, characterized in that, The threshold correction module includes: A coupling analysis unit that analyzes the mechanical vibration component in the vehicle acceleration data and extracts the characteristics of the electromagnetic pulse interference frequency band; A gene chain encoding unit that inputs the mechanical vibration component and the interference pattern characteristics in the historical false alarm event library into a genetic algorithm compensation model to generate an anti-interference compensation coefficient gene chain; A trend verification unit that uses a sliding window mechanism to perform continuous frame data trend matching verification on the safety height threshold corrected by the anti-interference compensation coefficient.

6. The vehicle detection system for a booster station according to claim 1, wherein, The closed-loop feedback module includes: A fingerprint modeling unit that receives the threshold trigger flag bit and device operation status data output by the threshold correction module, correlates and models the electromagnetic interference type and spatial position, and generates a feature map including the spatial coordinates of the interference source; A rule evolution engine that encodes the interference source coordinate data in the feature map and the disposal records in the historical false alarm event library into a genetic rule evolution model to generate an adaptable alarm logic rule set; A feedback adaptation unit that reversely transmits the actual disposal effect data of the alarm decision instruction to the fitness evaluation unit of the genetic algorithm spatial layout model of the environmental perception optimization module for optimizing the iterative parameters of the sensor topology gene coding sequence.

7. The vehicle detection system for a booster station according to claim 1, wherein, It also includes a data flow control module for: Establishing a real-time transmission link for electromagnetic field strength monitoring data from the electromagnetic spectrum acquisition unit of the environmental perception optimization module to the feature recognition layer of the joint denoising module; Configuring a synchronous update channel for the sensor weight parameters optimized by the genetic algorithm spatial layout model between the spatio-temporal calibration engine of the data fusion module and the gene chain encoding unit of the threshold correction module; Controlling the spatio-temporal correlation transmission path of vehicle motion acceleration data from the output end of the joint denoising module to the motion compensation unit of the data fusion module and the coupling analysis unit of the threshold correction module; Managing the reverse return channel of the disposal effect data generated by the feedback adaptation unit of the closed-loop feedback module to the fitness evaluation unit of the environmental perception optimization module to form a closed-loop optimization link for genetic algorithm parameters.

Images