Charging pile gun state detection method and system

Abstract

The invention relates to a charging pile gun state intelligent detection method and system, and the method comprises the steps: collecting a charging gun head electromagnetic signal, a gun handle temperature gradient and an interface contact surface mechanical strain signal in real time based on a multi-mode sensor array, constructing a space-time coupling data cube, separating the independent components of an electromagnetic field, a temperature field and a strain field, and obtaining a space-time coupling data cube; the multi-physical field coupling interference is eliminated; designing a dynamic cross compensation mechanism to correct mutual influence among sensor data; extracting multi-domain features and carrying out adversarial weight fusion to generate a comprehensive state representation vector; state scores are generated in combination with similarity matching of the dynamic reference library, and abnormal grade classification is achieved; according to the method, the problems of multi-field coupling interference separation, dynamic drift compensation and multi-dimensional feature joint modeling can be solved, the accuracy and real-time performance of poor contact of the charging gun, overheating early warning and mechanical deformation detection are improved, and a guarantee is provided for safe operation of charging facilities.

Description

Technical Field

[0001] The present invention relates to the technical field of charging equipment detection, and in particular to a charging pile gun status detection method and system. Background Art

[0002] With the rapid development of electric vehicles, the safety and reliability of charging piles, as core infrastructure, have attracted widespread attention. Charging guns, the key connecting component between charging piles and vehicles, are susceptible to poor contact, overheating, or deformation due to long-term plugging and unplugging, changes in ambient temperature and humidity, and mechanical stress. These issues can lead to safety hazards such as charging interruptions and fires. Traditional charging gun status detection relies primarily on monitoring single physical quantities such as temperature or current, or simple threshold alarm mechanisms based on multiple sensors. However, in actual operation, charging guns face the coupling of multiple physical fields, such as electromagnetic interference, abnormal thermal conductivity, and mechanical deformation. Single-dimensional detection methods cannot fully reflect the actual status, let alone accurately distinguish the correlation characteristics between normal operating conditions and potential faults.

[0003] While some existing methods attempt to fuse multi-sensor data, significant deficiencies remain in separating the coupled interference of multiple physical fields, compensating for dynamic cross-interference, and jointly analyzing multi-dimensional features in complex environments. For example, electromagnetic signals are susceptible to temperature drift and mechanical vibration, leading to distorted magnetic field strength measurements; temperature sensors lack reading stability in strong electromagnetic environments; and mechanical strain data often cannot strip away baseline drift caused by thermal expansion. Furthermore, existing methods lack modeling of the spatiotemporal correlations between multi-source data during feature extraction, making it difficult to establish highly robust state assessment models. Summary of the Invention

[0004] Based on this, the purpose of the present invention is to provide a charging pile gun status detection method and system that can achieve dynamic decoupling of multiple physical fields, accurately compensate for cross-interference, and effectively integrate multi-dimensional features.

[0005] The purpose of the present invention is achieved by the following scheme:

[0006] In a first aspect, the present invention provides a method for detecting the status of a charging pile gun, comprising the following steps:

[0007] S1: Based on the multimodal sensor array deployed at the interface between the charging gun tip and the gun handle, the charging gun is collected in real time to generate a charging sensor dataset containing the electromagnetic signal of the gun tip, the temperature signal of the gun handle, and the mechanical strain signal of the interface contact surface;

[0008] S2: Based on the charging sensor data set, a spatiotemporal coupling data cube is constructed, and the modal decomposition method is used to separate the electromagnetic field component, temperature field component, and strain field component of the spatiotemporal coupling data cube to generate electromagnetic sensing data, temperature sensing data, and strain sensing data;

[0009] S3: Perform cross-interference compensation processing on electromagnetic sensing data, temperature sensing data, and strain sensing data, and generate compensated electromagnetic data, compensated temperature data, and compensated strain data through dynamic coupling coefficient mapping and baseline drift calibration;

[0010] S4: performing multi-domain feature extraction and weighted fusion on the compensated electromagnetic data, compensated temperature data, and compensated strain data to generate a fused feature vector containing an electromagnetic feature vector, a temperature feature vector, and a strain feature vector;

[0011] S5: Perform similarity matching between the fused feature vector and the benchmark feature vector in the preset dynamic benchmark library to generate a status grade score representing the health status of the charging gun;

[0012] S6: Process the level status score based on the preset charging control rules, generate charging control instructions, and update the benchmark feature vector of the dynamic benchmark library based on the charging control instructions and the fused feature vector. The charging control instructions are used to control the start and stop of the charging pile or change the power operation.

[0013] In one embodiment, S1 of a charging pile gun status detection method provided by the present invention specifically includes the following steps:

[0014] S11: Perform high-frequency sampling and processing on the electromagnetic signals at the interface between the charging gun tip and the gun handle based on the magnetic field sensor array, capture the changes in the alternating magnetic field intensity, and generate an electromagnetic signal sequence;

[0015] S12: Using the temperature sensor, the temperature signal of the contact surface of the charging gun handle is spatially measured and processed to generate temperature gradient data. The temperature gradient data is used to identify heat conduction anomalies.

[0016] S13: Performing three-dimensional deformation measurement processing on the mechanical strain signal at the interface between the gun tip and the vehicle body based on the strain sensor array to generate a strain tensor field, which is used to quantify the mechanical stress distribution;

[0017] S14: Processing the electromagnetic signal sequence, the temperature gradient data, and the strain tensor field to generate a charging sensing data set containing electromagnetic signals, temperature signals, and mechanical strain signals.

[0018] In one embodiment, S2 of a charging pile gun status detection method provided by the present invention specifically includes the following steps:

[0019] S21: performing time window segmentation processing on the electromagnetic signal sequence of the charging sensor data set, extracting time domain segments according to the sliding window mechanism, and generating an electromagnetic signal time matrix;

[0020] S22: performing spatial normalization processing on the temperature gradient matrix of the charging sensor data set, mapping it to the grid nodes in the unified coordinate system, and generating a temperature gradient spatial matrix;

[0021] S23: performing node expansion processing on the strain tensor field of the charging sensor data set, reorganizing the data dimension according to the sensor position index, and generating a strain tensor node matrix;

[0022] S24: Perform three-dimensional stacking of the electromagnetic signal time matrix, temperature gradient space matrix, and strain tensor node matrix, integrate the data along the time, space, and physical field dimensions, and construct a spatiotemporal coupled data cube;

[0023] S25: Perform multilinear tensor decomposition on the spatiotemporal coupling data cube, separate the core tensor and factor matrix through the alternating least squares algorithm, and generate electromagnetic sensing data, temperature sensing data, and strain sensing data.

[0024] Furthermore, the multilinear tensor decomposition process is implemented by the following formula:

[0025]

[0026] in, It is a space-time coupled data cube, containing the original data of electromagnetic, temperature and strain fields. is the core tensor, U (t) 、U (s) 、U (m) are the factor matrices of time, space, and physical field dimensions, × n is the tensor n-module product, that is, the matrix multiplication is expanded along the n-th dimension.

[0027] In one embodiment, S3 of a charging pile gun status detection method provided by the present invention specifically includes the following steps:

[0028] S31: performing temperature-strain dynamic coupling compensation processing on the electromagnetic sensing data, adjusting the magnetic field gain parameter in real time, and generating compensated electromagnetic data. The compensated electromagnetic data is used to suppress the offset of the magnetic measurement signal caused by thermal expansion.

[0029] S32: Performing electromagnetic gradient compensation on the temperature sensing data, calculating the induced temperature drift based on the spatial gradient modulus of the magnetic field, and reversely correcting the original temperature reading to generate compensated temperature data. The compensated temperature data is used to eliminate the interference of the alternating magnetic field on the temperature measurement of the thermocouple;

[0030] S33: Performing thermal expansion baseline calibration on the strain sensing data, dynamically fitting the strain baseline drift curve through the temperature sensing data, and generating compensated strain data. The compensated strain data is used to remove the deformation measurement error caused by temperature change.

[0031] In one embodiment, S4 of a charging pile gun status detection method provided by the present invention specifically includes the following steps:

[0032] S41: performing multi-scale wavelet decomposition processing on the compensated electromagnetic data, extracting the energy norm of the high-frequency detail coefficient, and generating an electromagnetic eigenvector. The electromagnetic eigenvector is used to characterize the high-frequency oscillation abnormality caused by poor contact;

[0033] S42: performing short-time Fourier transform processing on the compensated temperature data, calculating the energy proportion of a preset fault frequency band, and generating a temperature characteristic vector. The temperature characteristic vector is used to indicate the thermal fluctuation characteristics caused by the abnormal contact resistance;

[0034] S43: performing empirical mode decomposition on the compensated strain data, extracting the instantaneous frequency variance of the intrinsic mode function, and generating a strain eigenvector. The strain eigenvector is used to detect mechanical jamming or deformation accumulation events.

[0035] S44: performing adversarial weight fusion processing on the electromagnetic feature vector, the temperature feature vector, and the strain feature vector, allocating weight coefficients of each feature dimension through a pre-trained network, and generating a fused feature vector containing the electromagnetic feature vector, the temperature feature vector, and the strain feature vector.

[0036] In one embodiment, S5 of a charging pile gun status detection method provided by the present invention specifically includes the following steps:

[0037] S51: Perform noise frequency band masking on the fused feature vector, shield the environmental vibration interference through a preset frequency band filter, and generate a denoised feature vector;

[0038] S52: Calculate the cosine similarity between the denoised feature vector and the benchmark feature vector in the preset dynamic benchmark library, and generate an initial state score by vector inner product and norm normalization. The calculation formula of the initial state score is:

[0039]

[0040] Among them, S is the initial state score, F is the denoising feature vector, and F ref is the benchmark feature vector in the dynamic benchmark library;

[0041] S53: The initial status score is graded based on a preset safety threshold, and divided into three levels: normal, warning, and fault according to a preset interval, to generate a status grade score representing the health status of the charging gun.

[0042] In a second aspect, the present invention provides a charging pile gun status detection system, which is configured with the following modules:

[0043] The charging data integration module is used to collect real-time data from the charging gun based on the multimodal sensor array deployed at the charging gun tip and gun handle interface, generating a charging sensor data set containing the electromagnetic signal of the gun tip, the temperature signal of the gun handle, and the mechanical strain signal of the interface contact surface;

[0044] The sensor data decomposition module is used to construct a spatiotemporal coupling data cube based on the charging sensor data set, and perform data separation on the spatiotemporal coupling data cube using the modal decomposition method to separate the electromagnetic field component, temperature field component, and strain field component to generate electromagnetic sensor data, temperature sensor data, and strain sensor data;

[0045] The charging interference compensation module is used to perform cross-interference compensation processing on electromagnetic sensing data, temperature sensing data and strain sensing data, and generates compensated electromagnetic data, compensated temperature data and compensated strain data through dynamic coupling coefficient mapping and baseline drift calibration;

[0046] The feature extraction and fusion module is used to perform multi-domain feature extraction and weighted fusion on the compensated electromagnetic data, compensated temperature data and compensated strain data to generate a fused feature vector containing the electromagnetic feature vector, the temperature feature vector and the strain feature vector;

[0047] The charging status scoring module is used to perform similarity matching between the fused feature vector and the benchmark feature vector in the preset dynamic benchmark library to generate a status grade score that represents the health status of the charging gun;

[0048] The charging control and benchmark update module is used to process the level status score based on the preset charging control rules, generate charging control instructions, and update the benchmark feature vector of the dynamic benchmark library based on the charging control instructions and the fused feature vector. The charging control instructions are used to control the start and stop of the charging pile or change the power operation.

[0049] In a third aspect, the present application provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements any of the above-mentioned charging pile gun status detection methods when executing the computer program.

[0050] In a fourth aspect, the present application provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, any of the above-mentioned charging pile gun status detection methods is implemented.

[0051] In summary, the charging pile gun status detection method provided by the present invention can effectively break through the limitations of traditional detection methods through multi-dimensional technology collaborative innovation, and significantly improve the accuracy and reliability of charging gun status assessment under complex working conditions. Real-time data acquisition based on a multimodal sensor array enables the simultaneous acquisition and spatiotemporal correlation modeling of electromagnetic, temperature, and mechanical strain signals, addressing the technical shortcoming of single-field monitoring that cannot cover the coupling effects of multiple fields. By constructing a spatiotemporal coupling data cube and applying a tensor decomposition algorithm, it achieves precise separation of multi-field interference and reconstruction of independent physical field components, effectively eliminating cross-interference issues such as thermally induced drift in the electromagnetic field, electromagnetic induction noise in the temperature field, and baseline offset due to thermal expansion in the strain field. The design of a dynamic coupling coefficient mapping and baseline drift calibration mechanism dynamically compensates for and suppresses the mutual influence between sensor data, improving the independence and measurement accuracy of each physical field data. The combination of multi-domain feature extraction and adversarial weight fusion technology constructs a highly robust multidimensional state representation vector for collaborative identification and joint analysis of fault characteristics such as poor contact, overheating anomalies, and mechanical deformation. A similarity matching and online update mechanism based on a dynamic benchmark library enables continuous optimization of the device's lifecycle condition assessment model, adapting to complex operating conditions such as charging gun aging and ambient temperature and humidity fluctuations, enabling early warning and graded response to abnormal conditions.

[0052] Through the organic integration of the above-mentioned technology chain, this method can comprehensively improve the intelligence level of charging pile gun connection reliability monitoring without relying on specific numerical thresholds, providing multi-dimensional and adaptive technical guarantees for the safe operation of charging facilities.

[0053] For better understanding and implementation, the present invention is described in detail below with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 A flow chart of a charging pile gun status detection method provided in an embodiment of the present application;

[0055] Figure 2 A schematic diagram of a process for generating electromagnetic sensing data, temperature sensing data, and strain sensing data provided in an embodiment of the present application;

[0056] Figure 3 A schematic diagram of a process for generating a status grade score according to an embodiment of the present application;

[0057] Figure 4 This is a structural diagram of a charging pile gun status detection system provided in another embodiment of the present application. DETAILED DESCRIPTION

[0058] To facilitate understanding of the present invention, the present invention will be described more fully below with reference to the accompanying drawings. The drawings illustrate preferred embodiments of the present invention. However, the present invention may be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to provide a more thorough and comprehensive understanding of the disclosure.

[0059] Unless otherwise defined, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which this invention pertains. The terms used in this specification are for the purpose of describing specific embodiments only and are not intended to limit the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0060] In one embodiment, Figure 1 As shown, a method for detecting the status of a charging pile gun is provided. This embodiment uses the method applied to a terminal as an example for illustration. It is understandable that the method can also be applied to a server, or to a system including a terminal and a server, and implemented through the interaction between the terminal and the server. In this embodiment, the method includes the following steps:

[0061] S1: Based on the multimodal sensor array deployed at the interface between the charging gun tip and the gun handle, the charging gun is collected in real time to generate a charging sensor dataset containing the electromagnetic signal of the gun tip, the temperature signal of the gun handle, and the mechanical strain signal of the interface contact surface.

[0062] Specifically, the multimodal sensor array consists of electromagnetic sensors, temperature sensors, and mechanical strain sensors, which respectively collect electromagnetic signals from the gun tip, temperature signals from the gun handle, and mechanical strain signals from the interface contact surface. These sensors offer high precision, high sampling rates, and excellent anti-interference performance, ensuring the accuracy and reliability of the collected data.

[0063] During the charging process, electromagnetic sensors monitor the electromagnetic field in the gun tip area in real time, capturing subtle changes in electromagnetic parameters such as magnetic and electric field strength. A temperature sensor, tightly fitted to the gun handle, senses temperature changes in this area in real time, reflecting thermal effects caused by factors such as the flow of current during charging. Mechanical strain sensors, precisely positioned at the interface, sensitively capture deformation caused by insertion and removal, mechanical stress, and other factors.

[0064] When the charging gun starts working, each sensor starts working synchronously according to the preset sampling frequency, and digitizes the collected electromagnetic signals, temperature signals, and mechanical strain signals. These digitized signals are transmitted to the system's data processing unit via the data transmission line and integrated according to the pre-set data structure and time series to finally generate a complete charging sensor data set. This data set contains multi-dimensional information such as electromagnetic, temperature, and mechanical strain of the charging gun during operation, providing basic data support for subsequent data analysis and processing. During the data collection process, the system also performs preliminary filtering and denoising on the data to remove possible environmental interference signals and ensure that the data quality meets the requirements of subsequent processing.

[0065] S2: Based on the charging sensor data set, a spatiotemporal coupling data cube is constructed, and the modal decomposition method is used to separate the electromagnetic field components, temperature field components, and strain field components to generate electromagnetic sensing data, temperature sensing data, and strain sensing data.

[0066] Specifically, the system constructs a spatiotemporal coupled data cube based on this dataset. This data cube is the key structure for the system's comprehensive analysis of the charging gun's operating status. Its three dimensions correspond to time, spatial location, and physical modality. By systematically arranging and integrating multimodal sensor data at different time points and spatial locations, a three-dimensional data matrix is formed, the spatiotemporal coupled data cube. This data structure comprehensively describes the changes in the charging gun's physical state at different time and spatial locations, providing strong data support for subsequent precise analysis.

[0067] Modal decomposition is an effective signal processing technique that can decompose complex mixed signals into several independent intrinsic mode functions, each of which corresponds to a specific physical modal component. In this embodiment, the system can use modal decomposition algorithms such as empirical mode decomposition (EMD) or variational mode decomposition (VMD). By performing modal decomposition on a spatiotemporal coupled data cube, the system can separate the originally coupled electromagnetic, temperature, and strain signals, extracting the electromagnetic field component, temperature field component, and strain field component, respectively.

[0068] During the modal decomposition process, the system identifies and selects the decomposed modal components based on the actual operating characteristics and physical laws of the charging gun. By analyzing the frequency characteristics and energy distribution of each modal component, the system can accurately determine which modal components correspond to the electromagnetic field, temperature field, and strain field components. After separating the electromagnetic field components, the system organizes them into electromagnetic sensing data. This dataset contains detailed information about the electromagnetic field changes in the gun tip area during the operation of the charging gun, such as the temporal variation of the magnetic field intensity and the spatial distribution characteristics of the electromagnetic field. Similarly, the temperature field components are organized into temperature sensing data, which reflects the dynamic temperature changes in the gun handle area, including important information such as the temperature rise rate, the time and location of the temperature peak. The strain field components are converted into strain sensing data, which details the mechanical strain at the interface contact surface, such as the strain variation curve over time and the strain distribution pattern.

[0069] After the aforementioned modal decomposition and data separation, the system generates electromagnetic, temperature, and strain sensing data representing electromagnetic, temperature, and strain characteristics, respectively. This separated data plays a vital role in subsequent analysis, providing a solid foundation for accurate assessment of the charging gun's status.

[0070] S3: Perform cross-interference compensation processing on the electromagnetic sensing data, temperature sensing data, and strain sensing data, and generate compensated electromagnetic data, compensated temperature data, and compensated strain data through dynamic coupling coefficient mapping and baseline drift calibration.

[0071] Specifically, the system performs cross-interference compensation on the separated electromagnetic, temperature, and strain sensor data to improve data purity and accuracy. For electromagnetic sensor data, considering that electromagnetic signals are susceptible to temperature drift and mechanical vibration, a dynamic coupling coefficient mapping method is used to establish coupling relationship models between electromagnetic and temperature, and electromagnetic and mechanical strain. By monitoring changes in temperature and mechanical strain in real time and utilizing pre-calibrated coupling coefficients, the electromagnetic sensor data is dynamically compensated to eliminate electromagnetic signal distortion caused by temperature drift and mechanical vibration. For temperature sensor data, due to its limited reading stability in strong electromagnetic environments, an electromagnetic field interference correction algorithm is employed to analyze the interference characteristics of the electromagnetic field on the temperature sensor and correct the temperature reading to ensure accurate temperature measurement. For strain sensor data, baseline drift caused by thermal expansion is a concern. A thermal expansion baseline drift correction algorithm, combined with real-time temperature data and the material thermal expansion coefficient, is used to perform baseline correction on the mechanical strain data, isolating the true mechanical strain information. After this cross-interference compensation process, compensated electromagnetic data, compensated temperature data, and compensated strain data are generated, providing a high-quality data source for subsequent feature extraction.

[0072] S4: Perform multi-domain feature extraction and weighted fusion on the compensated electromagnetic data, compensated temperature data, and compensated strain data to generate a fused feature vector containing an electromagnetic feature vector, a temperature feature vector, and a strain feature vector.

[0073] Specifically, multi-domain feature extraction is performed on the compensated electromagnetic data, temperature data, and strain data to explore key data features in different domains, including the time domain, frequency domain, and time-frequency domain. For electromagnetic data, statistical features such as mean, variance, and peak value are extracted in the time domain; spectral features such as frequency components and amplitude and phase are extracted in the frequency domain; and features such as energy distribution and time-frequency correlation are extracted in the time-frequency domain. Feature extraction for temperature data focuses on features such as the temperature change rate and temperature rise in the time domain; thermal periodicity in the frequency domain; and thermal diffusion characteristics in the time-frequency domain. For strain data, features such as strain amplitude and strain change rate in the time domain, vibration frequency components in the frequency domain, and strain energy distribution in the time-frequency domain are extracted. After extracting the eigenvectors of each physical mode, a weighted fusion algorithm is used to assign corresponding weights to different eigenvectors. Based on the importance of each feature in reflecting the charging gun status, the electromagnetic, temperature, and strain eigenvectors are weighted and fused together to generate a comprehensive fused feature vector that comprehensively and accurately represents the operating status of the charging gun at the current moment.

[0074] S5: Perform similarity matching on the fused feature vector and the benchmark feature vector in the preset dynamic benchmark library to generate a status grade score representing the health status of the charging gun.

[0075] Specifically, the benchmark feature vectors in the dynamic benchmark library are pre-established based on a large amount of operating data of the charging gun under normal operating conditions and known fault conditions, and contain typical feature patterns under different health states. Preferably, the system can use a similarity matching algorithm, such as Euclidean distance, cosine similarity, etc., to calculate the similarity value between the fused feature vector and the benchmark feature vector. According to the size of the similarity value, combined with the preset similarity threshold range, a status level score representing the health status of the charging gun is generated. The status level score can be represented by a quantitative numerical value, such as 0-100 points, where a higher score indicates that the charging gun is in a healthy operating state and close to the benchmark characteristics; a lower score indicates that the charging gun may have a fault risk and is significantly different from the benchmark characteristics. By matching and analyzing the fused feature vector with the dynamic benchmark library, a quantitative assessment of the health status of the charging gun is achieved, providing a basis for subsequent charging control decisions.

[0076] S6: Process the level status score based on the preset charging control rules, generate charging control instructions, and update the benchmark feature vector of the dynamic benchmark library based on the charging control instructions and the fused feature vector. The charging control instructions are used to control the start and stop of the charging pile or change the power operation.

[0077] Specifically, the preset charging control rules typically include specific control strategies corresponding to different scoring intervals. For example, when the status level score is higher than a certain set value, the charging pile is allowed to operate normally and the charging power is adjusted according to actual needs. When the score is lower than a certain threshold, an alarm is issued and the charging pile is controlled to suspend charging or reduce the charging power to ensure the safety of the charging process. After generating the corresponding charging control instructions, in addition to sending the control instructions to the charging pile to execute the start-stop or power operation change operation, the control instructions and the fused feature vector are also used to update the baseline feature vector of the dynamic benchmark library. By continuously combining new operating data and control decision information, the feature vectors in the benchmark library are dynamically adjusted, allowing the dynamic benchmark library to adapt to changes in the operating status of the charging gun, maintaining the accuracy and timeliness of the charging gun status assessment, and thus achieving continuous and effective monitoring and control of the charging pile gun status.

[0078] Preferably, after completing the execution of the charging control instruction, the system can also synchronously generate a charging abnormality status report containing the abnormality type, occurrence time, influencing parameters and recommended measures, and push the report to the user's mobile terminal device in real time through the Internet of Things communication module of the charging pile. The user can view the real-time working parameters, historical status trajectory and abnormal diagnosis details of the charging pile through the terminal application interface; at the same time, the user terminal establishes a two-way communication link with the sensor network and edge processor of the charging pile through an encrypted network hotspot to support the user to remotely adjust the charging power threshold, set safety protection strategy or trigger emergency stop command based on the abnormality report, so as to realize dynamic intervention and risk closed-loop management of the charging process, thereby improving the user's transparent monitoring capability and active control authority of the charging status on the basis of ensuring charging safety, and forming a full-link safety protection system of "abnormal perception-intelligent decision-making-user collaboration".

[0079] In summary, the charging pile gun status detection method provided by the present invention can effectively break through the limitations of traditional detection methods through multi-dimensional technology collaborative innovation, and significantly improve the accuracy and reliability of charging gun status assessment under complex working conditions. Real-time data acquisition based on a multimodal sensor array enables the simultaneous acquisition and spatiotemporal correlation modeling of electromagnetic, temperature, and mechanical strain signals, addressing the technical shortcoming of single-field monitoring that cannot cover the coupling effects of multiple fields. By constructing a spatiotemporal coupling data cube and applying a tensor decomposition algorithm, it achieves precise separation of multi-field interference and reconstruction of independent physical field components, effectively eliminating cross-interference issues such as thermally induced drift in the electromagnetic field, electromagnetic induction noise in the temperature field, and baseline offset due to thermal expansion in the strain field. The design of a dynamic coupling coefficient mapping and baseline drift calibration mechanism dynamically compensates for and suppresses the mutual influence between sensor data, improving the independence and measurement accuracy of each physical field data. The combination of multi-domain feature extraction and adversarial weight fusion technology constructs a highly robust multidimensional state representation vector for collaborative identification and joint analysis of fault characteristics such as poor contact, overheating anomalies, and mechanical deformation. A similarity matching and online update mechanism based on a dynamic benchmark library enables continuous optimization of the device's lifecycle condition assessment model, adapting to complex operating conditions such as charging gun aging and ambient temperature and humidity fluctuations, enabling early warning and graded response to abnormal conditions.

[0080] Through the organic integration of the above-mentioned technology chain, this method can comprehensively improve the intelligence level of charging pile gun connection reliability monitoring without relying on specific numerical thresholds, providing multi-dimensional and adaptive technical guarantees for the safe operation of charging facilities.

[0081] In one embodiment, S1 of a charging pile gun status detection method provided by the present invention specifically includes the following steps:

[0082] S11: Based on the magnetic field sensor array, high-frequency sampling and processing are performed on the electromagnetic signal at the interface between the charging gun head and the gun handle, the changes in the alternating magnetic field intensity are captured, and an electromagnetic signal sequence is generated.

[0083] Specifically, during the charging process, an alternating magnetic field is generated when current passes through the interface between the gun tip and the gun handle. The magnetic field sensor array samples this magnetic field in real time at an extremely high sampling frequency to ensure that no critical information is missed. The sampling frequency is set based on the operating characteristics of the charging gun and the changing patterns of the electromagnetic signal. It has undergone rigorous experimental verification and theoretical calculations to fully meet the needs of accurately capturing the high-frequency components of the electromagnetic signal. Through this high-frequency sampling process, the system can generate a detailed electromagnetic signal sequence that accurately reflects the dynamic changes in magnetic field intensity over time, including key characteristics such as the peak intensity of the magnetic field, the change period, and the fluctuation trend. These data provide high-resolution and high-precision basic information for subsequent analysis of the electromagnetic state of the charging gun, enabling the system to keenly perceive any abnormal changes in the electromagnetic field.

[0084] S12: Based on the temperature sensor, the temperature signal of the contact surface of the charging gun handle is spatially measured and processed to generate temperature gradient data. The temperature gradient data is used to identify heat conduction anomalies.

[0085] Specifically, during the operation of the charging gun, the temperature of the gun handle contact surface will exhibit certain spatial variations due to the thermal effects of the current and environmental factors. Temperature sensors are tightly fitted to different locations on the gun handle contact surface in a scientific and rational layout to ensure accurate acquisition of the temperature distribution in that area. These sensors can sense subtle temperature changes in real time and convert the measurement results into electrical signals for transmission. The system performs professional spatial distribution processing on the collected temperature signals and constructs temperature gradient data using advanced data processing algorithms. This temperature gradient data intuitively demonstrates the spatial temperature variation pattern of the gun handle contact surface, including important information such as the temperature rise rate, the direction of the temperature gradient, and the distribution range of high-temperature areas.

[0086] Through in-depth analysis of this temperature gradient data, the system can effectively identify thermal conductivity anomalies. For example, an abnormally high temperature or excessively large temperature gradient in a local area may indicate problems such as poor contact or uneven heat dissipation, providing a strong basis for system fault diagnosis and safety warnings.

[0087] S13: Perform three-dimensional deformation measurement processing on the mechanical strain signal of the interface contact surface between the gun tip and the vehicle body based on the strain sensor array to generate a strain tensor field, which is used to quantify the mechanical stress distribution.

[0088] Specifically, the system uses a strain sensor array to perform three-dimensional deformation measurement based on the mechanical strain signals at the interface between the charging connector tip and the vehicle body. This strain sensor array, composed of multiple highly sensitive strain gauges, can capture changes in mechanical strain from various directions. Factors such as plugging and unplugging the charging connector and vehicle vibration during driving cause complex mechanical deformations at the interface between the connector tip and the vehicle body. The strain sensor array senses these deformations in real time and converts them into electrical signals for output. The system employs a specialized three-dimensional deformation measurement algorithm to perform in-depth processing on the collected strain signals to generate a strain tensor field. This strain tensor field provides a detailed description of the mechanical stress distribution at the interface in three dimensions, encompassing multiple dimensions such as strain magnitude, direction, and trend. By analyzing the strain tensor field, the system can quantify the mechanical stress distribution and accurately assess the mechanical condition of the interface. This helps promptly identify issues such as loosening and deformation of the interface caused by excessive mechanical stress, providing critical data support for ensuring the reliability of the connection between the charging connector and the vehicle body.

[0089] S14: Processing the electromagnetic signal sequence, the temperature gradient data, and the strain tensor field to generate a charging sensing data set containing electromagnetic signals, temperature signals, and mechanical strain signals.

[0090] Specifically, the system synchronizes the time and formats of electromagnetic signal sequences, temperature gradient data, and strain tensor fields, ensuring that all data can be integrated within the same time frame. Through specialized data fusion algorithms, the system organically combines electromagnetic, temperature, and mechanical strain signals to form a comprehensive dataset containing information from multiple physical modalities. This charging sensor dataset not only contains a wealth of electromagnetic, temperature, and mechanical strain data, but also preserves the correlation between these data in terms of time series and spatial distribution. This provides a comprehensive, accurate, and highly relevant data foundation for subsequent comprehensive analysis of the charging gun's status. For example, when analyzing the overall performance of a charging gun, one can simultaneously refer to the electrical connection status reflected by the electromagnetic signal, the heat conduction conditions revealed by the temperature gradient data, and the mechanical stress distribution displayed by the strain tensor field, thereby more accurately assessing the overall operating status of the charging gun and providing solid data support for subsequent status assessment and fault diagnosis.

[0091] In one embodiment, Figure 2 As shown, S2 of a charging pile gun status detection method provided by the present invention specifically includes the following steps:

[0092] S21: Perform time window segmentation processing on the electromagnetic signal sequence of the charging sensor data set, extract time domain segments according to the sliding window mechanism, and generate an electromagnetic signal time matrix.

[0093] Specifically, the system sets an appropriate time window length and sliding step size based on the characteristics of the electromagnetic signal and the operating frequency of the charging gun. The time window length ensures complete coverage of the periodic characteristics of the electromagnetic signal, while the sliding step size balances data continuity and computational efficiency. For example, for high-frequency electromagnetic signals, the time window length may be set to milliseconds, with a sliding step size of half the window length to preserve subtle signal variations. In actual processing, the system first preprocesses the electromagnetic signal sequence to remove DC offsets and filter and remove noise to improve data quality. The preprocessed signal is segmented into multiple time domain segments, each representing the electromagnetic signal characteristics within a time window. The system arranges these segments in chronological order to form an electromagnetic signal time matrix, where rows represent time windows and columns represent signal sampling points. This matrix intuitively displays the temporal evolution of the electromagnetic signal, facilitating subsequent analysis of changes in the electromagnetic state of the charging gun, such as signal stability, periodic fluctuations, and abnormal mutations. This time window segmentation method enables the system to accurately capture the dynamic changes of the electromagnetic signal, providing a high-quality data foundation for subsequent electromagnetic field analysis.

[0094] S22: Perform spatial normalization on the temperature gradient matrix of the charging sensor data set, map it to the grid nodes in a unified coordinate system, and generate a temperature gradient spatial matrix.

[0095] Specifically, the system constructs a three-dimensional coordinate system to fully cover the measurement area of the charging gun's grip contact surface. Using interpolation algorithms such as inverse distance weighted (IDW) or kriging, the system maps the discrete temperature data collected by the temperature sensors onto a regular grid of nodes. The interpolation algorithm is selected based on the sensor layout and temperature field characteristics to ensure the accuracy of the temperature gradient. During normalization, the system eliminates data deviations caused by differences in sensor placement, ensuring consistent spatial distribution of temperature data. The rows of the temperature gradient spatial matrix correspond to spatial locations, and the columns correspond to temperature gradient components, clearly demonstrating the spatial variation of the temperature field. By analyzing this matrix, the system can identify thermal conduction anomalies, such as localized overheating or uneven heat dissipation. For example, if a region in the matrix shows a sharp temperature increase while surrounding areas show a more gradual temperature change, this may indicate poor contact or heat dissipation issues in that area. Through this spatial normalization process, the system provides a reliable basis for evaluating the thermal performance of the charging gun and ensuring thermal safety during the charging process.

[0096] S23: Perform node expansion processing on the strain tensor field of the charging sensor data set, reorganize the data dimension according to the sensor position index, and generate a strain tensor node matrix.

[0097] Specifically, the strain tensor field contains multiple strain components to comprehensively describe the three-dimensional deformation state of the interface contact surface. Specifically, the system reorganizes the strain tensor field data by sensor location based on the position index of the strain sensor at the interface between the charging gun tip and the vehicle body. Each sensor location corresponds to a row in the matrix, and the columns correspond to different strain components, such as longitudinal strain, transverse strain, and shear strain. Node expansion ensures that the strain data of each sensor is presented independently in the matrix, facilitating subsequent analysis of the mechanical stress distribution at each location. For example, a significant increase in longitudinal strain at a certain location in the matrix, while the strain at other locations shows little change, may indicate that the location is subject to significant mechanical tension. Through this reorganization method, the system can intuitively display the strain characteristics of each sensor location, providing detailed data support for evaluating the mechanical stability of the interface contact surface. Utilizing the strain tensor node matrix, the system can promptly detect deformation or loosening caused by mechanical stress, ensuring the reliability of the connection between the charging gun and the vehicle body.

[0098] S24: Perform three-dimensional stacking of the electromagnetic signal time matrix, temperature gradient space matrix, and strain tensor node matrix, integrate the data along the time, space, and physical field dimensions, and construct a space-time coupled data cube.

[0099] Specifically, during the 3D stacking process, the system first ensures the time series consistency of the three matrices, aligning the timestamps of the electromagnetic, temperature, and strain data using a time synchronization algorithm. In the spatial dimension, the system uses a unified coordinate system to map the spatial locations in the different matrices to the same grid nodes, ensuring spatial correspondence of the data. In the physical field dimension, the system distinguishes between the three physical modes of electromagnetic, temperature, and strain, assigning independent channels to each mode. The resulting integrated spatiotemporal coupled data cube is a three-dimensional array, where each element corresponds to a measurement value at a specific time, spatial location, and physical field. This data structure comprehensively describes the electromagnetic, temperature, and strain state changes of the charging gun during operation, reflecting the inter-physics coupling relationship. For example, the data cube shows coupled phenomena such as temperature rise caused by electromagnetic signal changes and mechanical strain increase caused by temperature changes. The spatiotemporal coupled data cube provides a unified data framework for subsequent multi-physics analysis, facilitating a deeper understanding of the complex operating state of the charging gun and laying the foundation for accurate condition assessment and fault diagnosis.

[0100] S25: Perform multilinear tensor decomposition on the spatiotemporal coupling data cube, separate the core tensor and factor matrix through the alternating least squares algorithm, and generate electromagnetic sensing data, temperature sensing data, and strain sensing data.

[0101] Specifically, multilinear tensor decomposition is a powerful high-dimensional data analysis method that can reveal hidden structures and patterns in the data. The alternating least squares algorithm uses iterative optimization to alternately update the core tensor and factor matrix to minimize reconstruction error. During the decomposition process, the system decomposes the spatiotemporal coupled data cube into a core tensor and three factor matrices, corresponding to the time, space, and physical field dimensions. The core tensor represents the strength of the interaction between the three dimensions, while the factor matrices describe the characteristic patterns along each dimension. For example, an element in the core tensor might represent the strength of the coupling between a specific time, spatial location, and physical field. By analyzing the core tensor and factor matrices, the system extracts independent features of the three physical modes: electromagnetic, temperature, and strain, and generates corresponding sensor data. For example, the core tensor can identify periodic characteristics of electromagnetic signals, thermal conduction patterns in the temperature field, and mechanical stress distribution patterns in the strain field. This decomposition method not only effectively separates multi-physics field data but also preserves the inherent connections between the data, facilitating subsequent analysis of the characteristics of each physical field and improving the accuracy and reliability of charging gun condition assessment. The system uses the multilinear tensor decomposition results to deeply understand the operating status of the charging gun, detect potential faults in a timely manner, and ensure the safety and stability of the charging process. Preferably, the multilinear tensor decomposition processing is implemented through the following formula:

[0102]

[0103] in, It is a space-time coupled data cube, containing the original data of electromagnetic, temperature and strain fields. is the core tensor, U (t) 、U (s) 、U (m) are the factor matrices of time, space, and physical field dimensions, × n is the tensor n-module product, that is, the matrix multiplication is expanded along the n-th dimension.

[0104] The above-mentioned charging pile gun status detection method can effectively solve the technical problem of traditional detection methods being difficult to accurately separate independent physical field information under the interference of multi-physical field coupling through multi-dimensional data modeling and tensor decomposition technology. Based on the time window segmentation mechanism, the time domain slicing processing of the electromagnetic signal sequence can realize the refined capture of the dynamic characteristics of the signal, which is used to construct a time matrix with a temporal evolution law; through the spatial normalization mapping and node index reorganization technology, the temperature gradient matrix and strain tensor field are standardized to achieve spatial alignment and dimensional consistency of cross-sensor data, which is used to eliminate data heterogeneity caused by differences in sensor layout; three-dimensional stacking processing technology can deeply integrate time, space and physical field dimensional information to construct a data cube with spatiotemporal correlation characteristics, which is used to realize the joint modeling and global feature expression of multi-source heterogeneous data; the multilinear tensor decomposition algorithm combined with the alternating least squares optimization strategy can separate the core tensor and orthogonal factor matrix from the coupled data, which is used to realize the independent analysis and interference stripping of electromagnetic field, temperature field and strain field data.

[0105] Through the synergistic effect of the above-mentioned technical processes, it is possible to break through the limitations of traditional threshold criteria in adapting to complex working conditions without pre-setting the physical field coupling model, significantly improve the accuracy of physical field data separation and the reliability of state feature extraction under multi-field coupling interference, and provide a high-fidelity multimodal data foundation for the health status assessment of charging guns.

[0106] In one embodiment, S3 of a charging pile gun status detection method provided by the present invention specifically includes the following steps:

[0107] S31: Performing temperature-strain dynamic coupling compensation processing on the electromagnetic sensing data, adjusting the magnetic field gain parameters in real time, and generating compensated electromagnetic data. The compensated electromagnetic data is used to suppress the offset of the magnetic measurement signal caused by thermal expansion.

[0108] Specifically, the dynamic coupling compensation process is based on the complex coupling relationship between the electromagnetic field and temperature and strain, thereby eliminating the interference of temperature and strain changes on the electromagnetic measurement results. In this embodiment, the thermal expansion effect will cause a slight change in the physical position of the magnetic sensor, resulting in an offset in the magnetic field measurement signal. At the same time, the strain change will affect the magnetic properties of the material, such as the magnetic permeability, and thus change the propagation and response of the electromagnetic signal. The system monitors the data of the temperature and strain sensors in real time, and uses the pre-established temperature-strain-electromagnetic field coupling model to calculate the adjustment amount of the magnetic field gain parameter. The model integrates multiple physical field parameters such as the thermal expansion coefficient of the magnetic material, the change law of magnetic permeability with temperature, and the influence coefficient of the strain corresponding to the magnetic field response.

[0109] During the compensation process, the system applies real-time adjusted magnetic field gain parameters to the electromagnetic sensor data, dynamically correcting the raw electromagnetic data. For example, if a temperature increase causes a decrease in the sensitivity of the magnetic field sensor, the system increases the magnetic field gain accordingly to restore accurate electromagnetic signal measurement. The compensated electromagnetic data effectively suppresses magnetic signal offset caused by thermal expansion, improving the accuracy and reliability of electromagnetic measurements and providing a more realistic data foundation for subsequent electromagnetic state analysis.

[0110] S32: The temperature sensing data is subjected to electromagnetic gradient compensation processing, the induced temperature drift is calculated based on the spatial gradient modulus of the magnetic field, the original temperature reading is reversely corrected, and the compensated temperature data is generated. The compensated temperature data is used to eliminate the temperature measurement interference of the alternating magnetic field on the thermocouple.

[0111] Specifically, since the charging gun operates in an alternating magnetic field environment, the temperature sensor will be affected by the electromagnetic induction effect, resulting in the output temperature signal containing interference components related to the magnetic field change, namely the induced temperature drift. In this embodiment, the system first performs a spatial gradient analysis on the electromagnetic sensing data to calculate the gradient modulus of the magnetic field in space. This process can be implemented through algorithms such as the differential method or Gaussian filtering, so that the rate of change of the magnetic field at different positions can be accurately reflected. Based on the correlation model between the spatial gradient modulus of the magnetic field and the characteristics of the temperature sensor, the system calculates the magnitude and direction of the induced temperature drift. This correlation model is established by combining experimental calibration and theoretical analysis, taking into account factors such as the material properties, geometric dimensions, and position of the temperature sensor in the magnetic field.

[0112] After determining the induced temperature drift, the system performs a reverse correction on the original temperature reading, subtracting the effect of the induced temperature drift from the raw temperature data. For example, if a temperature reading is falsely high due to a magnetic field gradient, the system lowers the temperature reading accordingly based on the calculated induced temperature drift, generating compensated temperature data closer to the true temperature value. This step eliminates interference from alternating magnetic fields on thermocouple temperature measurement, improving the accuracy and stability of temperature measurement and ensuring the reliability of temperature data for charging gun status assessment.

[0113] S33: Performing thermal expansion baseline calibration on the strain sensing data, dynamically fitting the strain baseline drift curve through the temperature sensing data, and generating compensated strain data. The compensated strain data is used to remove the deformation measurement error caused by temperature change.

[0114] Specifically, in the actual use of charging guns, temperature changes can cause the charging gun material to expand or contract, resulting in baseline drift in the strain sensor. This drift is not caused by actual mechanical deformation but is a direct result of temperature changes, which can seriously affect the accuracy of strain measurement. The system uses real-time temperature data monitored by the temperature sensor, combined with physical properties such as the thermal expansion coefficient of the material, to dynamically fit the drift curve of the strain baseline. The fitting process uses advanced mathematical methods such as polynomial fitting, exponential fitting, and Kalman filtering to adapt to the baseline drift characteristics under different temperature change patterns. By analyzing the quantitative relationship between temperature change and strain baseline drift, the system can accurately calculate the amount of baseline drift.

[0115] After obtaining the baseline drift curve, the system strips it from the raw strain sensor data to generate compensated strain data. This process effectively eliminates deformation measurement errors caused by temperature changes, allowing the strain data to more accurately reflect the actual mechanical deformation of the charging gun interface. For example, if the ambient temperature rises during the operation of the charging gun, causing the strain sensor baseline to rise, the system can restore the actual strain change by dynamically fitting the baseline drift curve and performing calibration, providing accurate data support for evaluating the mechanical connection reliability of the charging gun.

[0116] In one embodiment, S4 of a charging pile gun status detection method provided by the present invention specifically includes the following steps:

[0117] S41: Perform multi-scale wavelet decomposition on the compensated electromagnetic data, extract the energy norm of the high-frequency detail coefficient, and generate an electromagnetic eigenvector. The electromagnetic eigenvector is used to characterize the high-frequency oscillation abnormality caused by poor contact.

[0118] Specifically, multiscale wavelet decomposition is an efficient time-frequency analysis method that can decompose signals into approximate and detail components at different scales. The system selects appropriate wavelet basis functions and decomposition scales to ensure that the high-frequency oscillation characteristics of the electromagnetic signal are captured. During the decomposition process, the system performs a multi-level decomposition of the compensated electromagnetic data, generating a series of detail coefficients with different frequency bandwidths. These detail coefficients reflect the high-frequency variations of the electromagnetic signal at different time scales. To quantify these high-frequency variations, the system calculates the energy norm of the detail coefficients at each scale. This energy norm is calculated using methods such as square sum normalization and can characterize the energy distribution and intensity of the high-frequency signal. The resulting electromagnetic eigenvector, which incorporates energy norms at multiple scales, comprehensively reflects the high-frequency characteristics of the electromagnetic signal. High-frequency oscillation anomalies are often a typical sign of poor contact. The electromagnetic eigenvector can effectively characterize the electromagnetic signal changes caused by poor contact, providing critical electromagnetic characteristic information for subsequent condition assessment.

[0119] S42: Performing short-time Fourier transform processing on the compensated temperature data, calculating the energy proportion of a preset fault frequency band, and generating a temperature characteristic vector. The temperature characteristic vector is used to indicate the thermal fluctuation characteristics caused by the abnormal contact resistance.

[0120] Specifically, the short-time Fourier transform (SFT) can analyze the spectral characteristics of a signal over different time periods and is particularly well-suited for processing non-stationary temperature signals. The system divides the temperature data into multiple short-time windows and performs a Fourier transform on the data within each window to obtain its spectral distribution. The preset fault frequency band is predetermined based on the typical frequency range of thermal fluctuations associated with abnormal contact resistance, typically including low-frequency drift and medium-frequency fluctuation segments. The system calculates the energy contribution of the preset fault frequency band within the spectrum of each short-time window—that is, the ratio of the energy within that frequency band to the total energy. This metric reflects the thermal fluctuation characteristics of the temperature signal associated with abnormal contact resistance. The temperature eigenvector, composed of the energy contributions of multiple short-time windows, demonstrates the dynamic changes in the temperature signal over time. By analyzing the temperature eigenvector, the system can promptly detect temperature fluctuations caused by abnormal contact resistance, providing an important basis for assessing the thermal status of the charging gun.

[0121] S43: Performing empirical mode decomposition on the compensated strain data, extracting the instantaneous frequency variance of the intrinsic mode function, and generating a strain eigenvector. The strain eigenvector is used to detect mechanical jamming or deformation accumulation events.

[0122] Specifically, empirical mode decomposition can decompose complex non-stationary signals into multiple intrinsic mode functions, each of which represents the fluctuation characteristics of the signal within a specific frequency range. The system adaptively decomposes the compensated strain data to obtain a series of intrinsic mode functions. These functions are arranged from high to low frequency, corresponding to different frequency components in the strain signal. In order to capture the time-varying characteristics of the strain signal, the system calculates the instantaneous frequency variance of each intrinsic mode function. The instantaneous frequency variance is calculated by methods such as the Hilbert transform, which reflects the degree of frequency fluctuation of the intrinsic mode function over time. The strain eigenvector is composed of the instantaneous frequency variance of each eigenmode function and can comprehensively characterize the time-frequency characteristics of the strain signal. Mechanical jamming or deformation accumulation events usually cause significant changes in the frequency characteristics of the strain signal. The strain eigenvector can effectively detect these abnormal events and provide key features for evaluating the mechanical state of the charging gun.

[0123] S44: performing adversarial weight fusion processing on the electromagnetic feature vector, the temperature feature vector, and the strain feature vector, allocating weight coefficients of each feature dimension through a pre-trained network, and generating a fused feature vector containing the electromagnetic feature vector, the temperature feature vector, and the strain feature vector.

[0124] Specifically, adversarial weight fusion is an advanced data fusion strategy that automatically adjusts the weights of features based on their importance and relevance. The system uses a pre-trained deep neural network to learn and analyze electromagnetic, temperature, and strain feature vectors. The pre-trained network is trained with a large amount of historical data and annotated information to learn the contribution of different feature dimensions in characterizing the status of the charging gun. During the fusion process, the network automatically assigns weight coefficients to highlight the feature dimensions that are more important for evaluating the status of the charging gun, while suppressing the influence of noise and irrelevant features. The fused feature vector organically integrates electromagnetic, temperature, and strain features to form a comprehensive and integrated feature representation. It not only contains the high-frequency oscillation information in the electromagnetic feature vector and the thermal fluctuation characteristics in the temperature feature vector, but also integrates the mechanical deformation event information in the strain feature vector, thereby more accurately reflecting the overall operating status of the charging gun and providing strong feature support for subsequent health status assessment and fault diagnosis.

[0125] In one embodiment, Figure 3 As shown, S5 of the charging pile gun status detection method provided by the present invention specifically includes the following steps:

[0126] S51: performing noise frequency band masking on the fused feature vector, shielding the environmental vibration interference through a preset frequency band filter, and generating a denoised feature vector.

[0127] Specifically, in the actual operating environment, the sensor data of the charging gun is inevitably affected by interference factors such as environmental vibration. These interferences will introduce noise frequency bands into the feature vector, affecting the accuracy of subsequent status assessment. Based on this, the system first performs spectral analysis on the fused feature vector to identify the noise frequency bands therein. The preset frequency band filter predetermines the main frequency band ranges of interference such as environmental vibration based on the actual working environment and historical data of the charging gun. During the processing process, the system applies these preset frequency band filters to the fused feature vector and performs mask processing on the identified noise frequency bands, that is, attenuates or shields the signals in these frequency bands, thereby effectively removing interference components such as environmental vibration. The denoised feature vector retains the effective features related to the status of the charging gun, improves the purity and representativeness of the feature vector, and provides a more reliable data basis for subsequent status assessment.

[0128] S52: performing cosine similarity calculation on the denoised feature vector and the benchmark feature vector in the preset dynamic benchmark library, and generating an initial state score by vector inner product and norm normalization.

[0129] Specifically, cosine similarity is a commonly used vector similarity measurement method that can measure the degree of similarity between two vectors in direction. The system first calculates the inner product of the denoised feature vector and the baseline feature vector to obtain their similarity measurement in the feature space. Then, the norms of the denoised feature vector and the baseline feature vector are calculated respectively, and normalized to eliminate the influence of the vector modulus difference on the similarity measurement. The final initial state score reflects the degree of similarity between the denoised feature vector and the baseline feature vector. The numerical value range is usually between -1 and 1. The closer the value is to 1, the higher the similarity. Preferably, the calculation formula for the initial state score is:

[0130]

[0131] Among them, S is the initial state score, F is the denoising feature vector, and F ref It is the benchmark feature vector in the dynamic benchmark library; this score provides a quantitative indicator of the similarity between the current state of the charging gun and the benchmark state, providing a basis for subsequent state level classification.

[0132] S53: The initial status score is graded based on a preset safety threshold, and divided into three levels: normal, warning, and fault according to a preset interval, to generate a status grade score representing the health status of the charging gun.

[0133] Specifically, preset safety thresholds are determined based on the charging gun's design specifications, operating experience, and safety standards, and serve as the boundary values for different health states. The system compares the initial health score with these safety thresholds and categorizes the health of the charging gun within a preset range. Typically, these ranges may be divided into three levels: normal, warning, and fault. For example, if the initial health score is above a certain upper threshold, the system determines that the charging gun is in a normal state, indicating that it is operating well and its characteristics are highly similar to the baseline state. If the score is below the upper threshold but above a lower threshold, the system issues a warning signal, indicating that there may be a minor anomaly or potential risk, requiring further attention and inspection. If the score is below the lower threshold, the system determines that the charging gun is in a faulty state, indicating that it has significant abnormal characteristics that may affect normal operation and requires timely repair or replacement. Through this grading process, the system generates a health rating score that represents the health of the charging gun, providing intuitive and clear status indicators for maintenance and management of the charging gun, helping to promptly identify and address potential issues and ensure safe and stable operation of the charging gun.

[0134] The above-mentioned charging pile gun status detection method can effectively solve the technical bottlenecks of traditional methods such as insufficient reliability of status scoring and delayed graded response under complex environmental interference through noise suppression and dynamic graded evaluation technology. The noise suppression mechanism based on frequency band mask processing can realize frequency domain feature selection and signal reconstruction of environmental vibration interference, so as to improve the anti-interference ability and representation purity of feature vectors; by projecting the reference feature vector in the vector space through cosine similarity calculation, a quantitative evaluation model of multi-dimensional state deviation can be constructed to realize dynamic calibration and continuous monitoring of the health status of charging guns; the graded processing technology combined with the adaptive threshold mapping strategy can achieve a refined interval division of normal operating conditions, potential risks and immediate failures, so as to realize early warning of abnormal conditions and graded response linkage.

[0135] The synergistic effect of the above-mentioned technology chain can break through the limitations of the traditional single scoring mechanism's lack of adaptability to complex working conditions without relying on fixed threshold criteria, significantly improve the environmental robustness of the status scoring and the timeliness of graded decision-making, and provide a highly reliable intelligent judgment basis for the precise execution of charging pile safety control strategies.

[0136] Preferably, if Figure 4 As shown, the present invention provides a charging pile gun status detection system 700, which is configured with the following modules:

[0137] Charging data integration module 710, used to collect real-time data from the charging gun based on the multimodal sensor array deployed at the charging gun tip and gun handle interface, generating a charging sensor data set containing electromagnetic signals from the gun tip, temperature signals from the gun handle, and mechanical strain signals from the interface contact surface;

[0138] A sensor data decomposition module 720 is configured to construct a spatiotemporal coupled data cube based on the charging sensor data set, and perform data separation on the spatiotemporal coupled data cube using a modal decomposition method to separate the electromagnetic field component, the temperature field component, and the strain field component to generate electromagnetic sensor data, temperature sensor data, and strain sensor data.

[0139] The charging interference compensation module 730 is used to perform cross-interference compensation processing on the electromagnetic sensing data, temperature sensing data, and strain sensing data, and generate compensated electromagnetic data, compensated temperature data, and compensated strain data through dynamic coupling coefficient mapping and baseline drift calibration;

[0140] The feature extraction and fusion module 740 is used to perform multi-domain feature extraction and weighted fusion on the compensated electromagnetic data, the compensated temperature data, and the compensated strain data to generate a fused feature vector containing an electromagnetic feature vector, a temperature feature vector, and a strain feature vector;

[0141] The charging status scoring module 750 is used to perform similarity matching between the fused feature vector and the reference feature vector in the preset dynamic reference library to generate a status grade score representing the health status of the charging gun;

[0142] The charging control and benchmark update module 760 is used to process the level status score based on the preset charging control rules, generate charging control instructions, and update the benchmark feature vector of the dynamic benchmark library based on the charging control instructions and the fused feature vector. The charging control instructions are used to control the start and stop of the charging pile or change the power operation.

[0143] In summary, the charging pile gun status detection system provided by the present invention can effectively break through the limitations of traditional detection methods through multi-dimensional technology collaborative innovation, and significantly improve the accuracy and reliability of charging gun status assessment under complex working conditions. Real-time data acquisition based on a multimodal sensor array enables the simultaneous acquisition and spatiotemporal correlation modeling of electromagnetic, temperature, and mechanical strain signals, addressing the technical shortcoming of single-field monitoring that cannot cover the coupling effects of multiple fields. By constructing a spatiotemporal coupling data cube and applying a tensor decomposition algorithm, it achieves precise separation of multi-field interference and reconstruction of independent physical field components, effectively eliminating cross-interference issues such as thermally induced drift in the electromagnetic field, electromagnetic induction noise in the temperature field, and baseline offset due to thermal expansion in the strain field. The design of a dynamic coupling coefficient mapping and baseline drift calibration mechanism dynamically compensates for and suppresses the mutual influence between sensor data, improving the independence and measurement accuracy of each physical field data. The combination of multi-domain feature extraction and adversarial weight fusion technology constructs a highly robust multidimensional state representation vector for collaborative identification and joint analysis of fault characteristics such as poor contact, overheating anomalies, and mechanical deformation. A similarity matching and online update mechanism based on a dynamic benchmark library enables continuous optimization of the device's lifecycle condition assessment model, adapting to complex operating conditions such as charging gun aging and ambient temperature and humidity fluctuations, enabling early warning and graded response to abnormal conditions.

[0144] Through the organic integration of the above-mentioned technology chain, this system can comprehensively improve the intelligence level of charging pile gun connection reliability monitoring without relying on specific numerical thresholds, providing multi-dimensional and adaptive technical guarantees for the safe operation of charging facilities.

[0145] Preferably, the charging data integration module 710 provided in this embodiment is configured with the following units:

[0146] The electromagnetic sequence generation unit is used to perform high-frequency sampling and processing on the electromagnetic signals at the interface between the charging gun head and the gun handle based on the magnetic field sensor array, capture the changes in the alternating magnetic field intensity, and generate an electromagnetic signal sequence;

[0147] A temperature gradient generation unit is used to perform spatial distribution measurement and processing of the temperature signal of the charging gun's handle contact surface based on the temperature sensor to generate temperature gradient data. The temperature gradient data is used to identify thermal conduction anomalies.

[0148] A strain tensor field generation unit is used to perform three-dimensional deformation measurement processing on the mechanical strain signal of the interface between the gun tip and the vehicle body based on the strain sensor array to generate a strain tensor field, which is used to quantify the mechanical stress distribution;

[0149] The charging sensor data integration unit is used to process the electromagnetic signal sequence, temperature gradient data and strain tensor field to generate a charging sensor data set containing electromagnetic signals, temperature signals and mechanical strain signals.

[0150] Preferably, the sensor data decomposition module 720 provided in this embodiment is configured with the following units:

[0151] An electromagnetic time matrix generation unit is used to perform time window segmentation processing on the electromagnetic signal sequence of the charging sensor data set, extract time domain segments according to the sliding window mechanism, and generate an electromagnetic signal time matrix;

[0152] A temperature space matrix generation unit is used to perform spatial normalization processing on the temperature gradient matrix of the charging sensor data set, map it to the grid nodes in a unified coordinate system, and generate a temperature gradient space matrix;

[0153] A strain node matrix generation unit is used to perform node expansion processing on the strain tensor field of the charging sensor data set, reorganize the data dimension according to the sensor position index, and generate a strain tensor node matrix;

[0154] The space-time coupling unit is used to perform three-dimensional stacking processing on the electromagnetic signal time matrix, temperature gradient space matrix and strain tensor node matrix, integrate data along the time, space and physical field dimensions, and construct a space-time coupling data cube;

[0155] The field component sensing data generation unit is used to perform multilinear tensor decomposition processing on the spatiotemporal coupling data cube, separate the core tensor and the factor matrix through the alternating least squares algorithm, and generate electromagnetic sensing data, temperature sensing data and strain sensing data.

[0156] Preferably, the charging interference compensation module 730 provided in this embodiment is configured with the following units:

[0157] The electromagnetic compensation unit is used to perform temperature-strain dynamic coupling compensation processing on the electromagnetic sensing data, adjust the magnetic field gain parameters in real time, and generate compensated electromagnetic data. The compensated electromagnetic data is used to suppress the offset of the magnetic measurement signal caused by thermal expansion;

[0158] The temperature compensation unit is used to perform electromagnetic gradient compensation on the temperature sensing data, calculate the induced temperature drift based on the spatial gradient modulus of the magnetic field, reversely correct the original temperature reading, and generate compensated temperature data. The compensated temperature data is used to eliminate the temperature measurement interference of the alternating magnetic field on the thermocouple;

[0159] The strain compensation unit is used to perform thermal expansion baseline calibration on the strain sensing data, dynamically fit the strain baseline drift curve through the temperature sensing data, and generate compensated strain data. The compensated strain data is used to strip off the deformation measurement error caused by temperature change.

[0160] Preferably, the feature extraction and fusion module 740 provided in this embodiment is configured with the following units:

[0161] An electromagnetic eigenvector generating unit is used to perform multi-scale wavelet decomposition on the compensated electromagnetic data, extract the energy norm of the high-frequency detail coefficient, and generate an electromagnetic eigenvector. The electromagnetic eigenvector is used to characterize the high-frequency oscillation anomaly caused by poor contact;

[0162] A temperature characteristic vector generation unit is used to perform short-time Fourier transform processing on the compensated temperature data, calculate the energy proportion of a preset fault frequency band, and generate a temperature characteristic vector. The temperature characteristic vector is used to indicate the thermal fluctuation characteristics caused by abnormal contact resistance;

[0163] A strain eigenvector generation unit is used to perform empirical mode decomposition on the compensated strain data, extract the instantaneous frequency variance of the intrinsic mode function, and generate a strain eigenvector. The strain eigenvector is used to detect mechanical jamming or deformation accumulation events.

[0164] The fusion feature vector generation unit is used to perform adversarial weight fusion processing on the electromagnetic feature vector, the temperature feature vector and the strain feature vector, and to assign weight coefficients of each feature dimension through a pre-trained network to generate a fusion feature vector containing the electromagnetic feature vector, the temperature feature vector and the strain feature vector.

[0165] Preferably, the charging status scoring module 750 provided in this embodiment is configured with the following units:

[0166] A denoising feature vector generating unit is used to perform noise frequency band masking on the fused feature vector, shield the environmental vibration interference through a preset frequency band filter, and generate a denoising feature vector;

[0167] An initial state score generating unit is used to perform cosine similarity calculation on the denoised feature vector and the benchmark feature vector in the preset dynamic benchmark library, and generate an initial state score by vector inner product and norm normalization;

[0168] The status grade score generation unit is used to grade the initial status score based on the preset safety threshold, divide it into three levels: normal, warning, and fault according to the preset interval, and generate a status grade score that represents the health status of the charging gun.

[0169] In one embodiment, the present application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor implements the above-mentioned charging pile gun status detection method when executing the computer program.

[0170] In one embodiment, the present application also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the above-mentioned charging pile gun status detection method is implemented.

[0171] In the description of this specification, the reference terms "one embodiment," "some embodiments," "example," "specific example," or "some examples" mean that the specific features, structures, materials, or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. Moreover, the specific features, structures, materials, or characteristics described may be combined in any appropriate manner in any one or more embodiments or examples. In addition, those skilled in the art may combine and integrate different embodiments or examples described in this specification, as well as features of different embodiments or examples, unless they are mutually inconsistent.

[0172] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to the partial description of the method embodiments. The device embodiments described above are merely illustrative, wherein the components described as separate parts may or may not be physically separated, and the parts displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the disclosed solution. A person of ordinary skill in the art can understand and implement it without expending creative work.

[0173] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any person skilled in the art can easily conceive of various modifications or substitutions within the technical scope disclosed in this application, and such modifications or substitutions should be included within the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.

Claims

1. A charging pile gun status detection method, characterized in that: The following steps are involved: S1: Based on the multimodal sensor array deployed at the interface between the charging gun tip and the gun handle, the charging gun is collected in real time to generate a charging sensor dataset containing the electromagnetic signal of the gun tip, the temperature signal of the gun handle, and the mechanical strain signal of the interface contact surface; S2: constructing a spatiotemporal coupling data cube based on the charging sensor data set, and performing data separation on the spatiotemporal coupling data cube by a modal decomposition method to separate electromagnetic field components, temperature field components, and strain field components to generate electromagnetic sensing data, temperature sensing data, and strain sensing data; S3: performing cross-interference compensation processing on the electromagnetic sensing data, temperature sensing data, and strain sensing data, and generating compensated electromagnetic data, compensated temperature data, and compensated strain data through dynamic coupling coefficient mapping and baseline drift calibration; S4: performing multi-domain feature extraction and weighted fusion on the compensated electromagnetic data, the compensated temperature data, and the compensated strain data to generate a fused feature vector containing an electromagnetic feature vector, a temperature feature vector, and a strain feature vector; S5: performing similarity matching on the fused feature vector and the reference feature vector in a preset dynamic reference library to generate a status grade score representing the health status of the charging gun; S6: Processing the grade status score based on preset charging control rules to generate a charging control instruction, and updating the benchmark feature vector of the dynamic benchmark library based on the charging control instruction and the fused feature vector. The charging control instruction is used to control the start and stop of the charging pile or change the power operation.

2. The method according to claim 1, characterized in that Said S1 comprises: S11: Perform high-frequency sampling and processing on the electromagnetic signals at the interface between the charging gun tip and the gun handle based on the magnetic field sensor array, capture the changes in the alternating magnetic field intensity, and generate an electromagnetic signal sequence; S12: performing spatial distribution measurement processing on the temperature signal of the contact surface of the charging gun handle using the temperature sensor to generate temperature gradient data, wherein the temperature gradient data is used to identify heat conduction anomalies; S13: performing three-dimensional deformation measurement processing on the mechanical strain signal of the interface contact surface between the gun tip and the vehicle body based on the strain sensor array to generate a strain tensor field, wherein the strain tensor field is used to quantify the mechanical stress distribution; S14: Processing the electromagnetic signal sequence, temperature gradient data, and strain tensor field to generate a charging sensing data set containing electromagnetic signals, temperature signals, and mechanical strain signals.

3. The method according to claim 1, characterized in that The S2 includes: S21: performing time window segmentation processing on the electromagnetic signal sequence of the charging sensor data set, extracting time domain segments according to a sliding window mechanism, and generating an electromagnetic signal time matrix; S22: performing spatial normalization processing on the temperature gradient matrix of the charging sensor data set, mapping the matrix to grid nodes in a unified coordinate system, and generating a temperature gradient spatial matrix; S23: performing node expansion processing on the strain tensor field of the charging sensor data set, reorganizing the data dimension according to the sensor position index, and generating a strain tensor node matrix; S24: performing three-dimensional stacking processing on the electromagnetic signal time matrix, the temperature gradient space matrix, and the strain tensor node matrix, integrating the data along the time, space, and physical field dimensions, and constructing a space-time coupled data cube; S25: performing multilinear tensor decomposition processing on the spatiotemporal coupling data cube, separating the core tensor and the factor matrix by an alternating least squares algorithm, and generating electromagnetic sensing data, temperature sensing data, and strain sensing data.

4. The method according to claim 3, characterized in that The multilinear tensor decomposition process is implemented by the following formula: in, It is a space-time coupled data cube, containing the original data of electromagnetic, temperature and strain fields. is the core tensor, U (t) 、U (s) 、U (m) are the factor matrices of time, space, and physical field dimensions, × n is the tensor n-module product, that is, the matrix multiplication is expanded along the n-th dimension.

5. The method according to claim 1, characterized in that The S3 includes: S31: performing temperature-strain dynamic coupling compensation processing on the electromagnetic sensing data, adjusting the magnetic field gain parameter in real time, and generating compensated electromagnetic data, wherein the compensated electromagnetic data is used to suppress the offset of the magnetic measurement signal caused by thermal expansion; S32: performing electromagnetic gradient compensation processing on the temperature sensing data, calculating the induced temperature drift based on the magnetic field spatial gradient modulus, reversely correcting the original temperature reading, and generating compensated temperature data, wherein the compensated temperature data is used to eliminate the interference of the alternating magnetic field on the temperature measurement of the thermocouple; S33: performing thermal expansion baseline calibration on the strain sensing data, dynamically fitting a strain baseline drift curve through the temperature sensing data, and generating compensated strain data, wherein the compensated strain data is used to remove deformation measurement errors caused by temperature changes.

6. The method according to claim 1, characterized in that The S4 includes: S41: performing multi-scale wavelet decomposition processing on the compensated electromagnetic data, extracting the energy norm of the high-frequency detail coefficient, and generating an electromagnetic feature vector, wherein the electromagnetic feature vector is used to characterize the high-frequency oscillation abnormality caused by poor contact; S42: performing short-time Fourier transform processing on the compensated temperature data, calculating the energy proportion of a preset fault frequency band, and generating a temperature characteristic vector, wherein the temperature characteristic vector is used to indicate thermal fluctuation characteristics caused by abnormal contact resistance; S43: performing empirical mode decomposition on the compensated strain data, extracting the instantaneous frequency variance of the intrinsic mode function, and generating a strain eigenvector, wherein the strain eigenvector is used to detect mechanical jamming or deformation accumulation events; S44: performing adversarial weight fusion processing on the electromagnetic feature vector, the temperature feature vector, and the strain feature vector, allocating weight coefficients of each feature dimension through a pre-trained network, and generating a fused feature vector containing the electromagnetic feature vector, the temperature feature vector, and the strain feature vector.

7. The method according to any one of claims 1 to 6, characterized in that The S5 includes: S51: performing noise frequency band masking processing on the fused feature vector, shielding environmental vibration interference through a preset frequency band filter, and generating a denoised feature vector; S52: Calculate the cosine similarity between the denoised feature vector and the reference feature vector in the preset dynamic reference library, and generate an initial state score by vector inner product and norm normalization. The calculation formula of the initial state score is: Among them, S is the initial state score, F is the denoising feature vector, and F ref is the benchmark feature vector in the dynamic benchmark library; S53: The initial status score is graded based on a preset safety threshold, and divided into three levels: normal, warning, and fault according to a preset interval, to generate a status grade score representing the health status of the charging gun.

8. A charging pile gun status detection system, characterized in that: The system comprises: The charging data integration module is used to collect real-time data from the charging gun based on the multimodal sensor array deployed at the charging gun tip and gun handle interface, generating a charging sensor data set containing the electromagnetic signal of the gun tip, the temperature signal of the gun handle, and the mechanical strain signal of the interface contact surface; a sensor data decomposition module, configured to construct a spatiotemporal coupled data cube based on the charging sensor data set, and perform data separation on the spatiotemporal coupled data cube using a modal decomposition method to separate electromagnetic field components, temperature field components, and strain field components, thereby generating electromagnetic sensor data, temperature sensor data, and strain sensor data; a charging interference compensation module, configured to perform cross-interference compensation processing on the electromagnetic sensing data, temperature sensing data, and strain sensing data, and generate compensated electromagnetic data, compensated temperature data, and compensated strain data through dynamic coupling coefficient mapping and baseline drift calibration; a feature extraction and fusion module, configured to perform multi-domain feature extraction and weighted fusion on the compensated electromagnetic data, the compensated temperature data, and the compensated strain data, to generate a fused feature vector containing an electromagnetic feature vector, a temperature feature vector, and a strain feature vector; A charging status scoring module is used to perform similarity matching between the fused feature vector and the reference feature vector in a preset dynamic reference library to generate a status grade score representing the health status of the charging gun; A charging control and benchmark update module is used to process the level status score based on preset charging control rules, generate charging control instructions, and update the benchmark feature vector of the dynamic benchmark library based on the charging control instructions and the fused feature vector. The charging control instructions are used to control the start and stop of the charging pile or change the power operation.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images