A direct current charging pile detection system and method applied to a site

Abstract

The application discloses a kind of direct current charging pile detection systems and methods applied to field, it is related to the online monitoring technical field of electric power equipment, including, three-dimensional data set is input Toeplitz cyclic measurement matrix and is compressed sampling, and high-resolution signal is reconstructed using improved orthogonal matching pursuit algorithm, local variation coefficient is analyzed and calculated by sliding window, and abnormal position is marked, and is integrated into abnormal position set;Abnormal position set is input pre-trained long short-term memory prediction model, assesses the remaining life of insulation and fault risk level, and real-time calculation direct current charging pile maximum allowable output current;Through memristor array dynamic adjustment detection load impedance, real-time acquisition direct current charging pile output current, and using synchronous compression wavelet transform analysis harmonic ridge line and extract harmonic component;Significantly improve the accuracy of fault early warning, and realize the rapid dynamic adjustment of output current, effectively solve the technical problem of fast charging pile multi-physical field coupling monitoring.

Description

A DC charging pile testing system and method for on-site application Technical Field

[0001] This invention relates to the field of online monitoring technology for power equipment, and in particular to a DC charging pile testing system and method for field application. Background Technology

[0002] In the field of DC charging pile technology, real-time status monitoring and fault early warning are core aspects of ensuring charging safety and equipment reliability. Distributed fiber optic sensors (DOFS) are used to synchronously collect strain and temperature data from the charging pile's output cable, and an anomaly alarm is triggered by a threshold trigger mechanism. By deploying multiple sensor nodes, a two-dimensional temperature-strain distribution model of the cable surface is constructed, and combined with preset safety thresholds, the initial location of the fault area is achieved. Furthermore, some improved schemes introduce Fourier transform or short-time-window (STFT) analysis of current harmonic components to assess the power quality of the charging pile's output. These methods perform well under steady-state conditions, and especially exhibit high detection sensitivity in scenarios with sudden changes in a single physical quantity (such as temperature).

[0003] Existing technologies still have limitations in multiphysics coupling analysis under dynamic operating conditions. Two-dimensional data models generated by distributed sensing systems struggle to characterize the three-dimensional degradation trajectory of cable insulation under complex stress-thermal cycling, leading to the easy neglect of local micro-strain accumulation or gradient anomalies (such as non-uniform deformation caused by mechanical vibration) by thresholding algorithms. Furthermore, traditional harmonic analysis methods, limited by the frequency domain resolution of fixed basis functions, struggle to capture the correlation between transient harmonic components and insulation aging. This limitation is particularly pronounced in the high current density scenarios of fast charging piles, necessitating a detection method that can integrate multi-dimensional dynamic characteristics and quantify insulation degradation risk. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a method for testing DC charging piles in the field to solve the problem of difficulty in accurately assessing the risk of cable insulation degradation.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] In a first aspect, the present invention provides a method for detecting DC charging piles in the field, which includes collecting the strain gradient and temperature gradient of the output cable of the DC charging pile, and performing preprocessing to generate a three-dimensional dataset.

[0008] The 3D dataset is input into the Toeplitz cyclic measurement matrix for compressed sampling, and then reconstructed into a high-resolution signal using an improved orthogonal matching pursuit algorithm. Local variation coefficients are calculated through sliding window analysis, anomaly locations are marked, and the results are integrated into an anomaly location set.

[0009] The abnormal location set is input into a pre-trained long short-term memory prediction model to assess the remaining insulation life and fault risk level, and to calculate the maximum allowable output current of the DC charging pile in real time.

[0010] The load impedance is dynamically adjusted by a memristor array, the output current of the DC charging pile is collected in real time, and the harmonic ridge is analyzed and the harmonic components are extracted by synchronous compressed wavelet transform.

[0011] By integrating the abnormal location set, fault risk level, and harmonic components, the power distribution network health index is output, and the optimal network reconfiguration scheme is generated through the quantum annealing algorithm.

[0012] As a preferred embodiment of the DC charging pile testing method applied in the field according to the present invention, the strain gradient and temperature gradient are time-aligned using GPS timestamps, and noise interference is eliminated by adaptive Kalman filtering and median filtering respectively. The filtered strain gradient and temperature gradient are interpolated to a unified spatiotemporal grid according to spatial location, time slice and physical quantity type to generate a three-dimensional dataset.

[0013] As a preferred embodiment of the DC charging pile detection method applied in the field according to the present invention, the specific steps of inputting the three-dimensional dataset into the Toeplitz cyclic measurement matrix for compressed sampling and reconstructing it into a high-resolution signal using an improved orthogonal matching pursuit algorithm are as follows.

[0014] The 3D dataset is divided into multiple sub-blocks according to spatial location, time slice, and physical quantity type;

[0015] The first row of elements is generated based on a random Gaussian distribution for each sub-block, and a Toeplitz cyclic measurement matrix is ​​generated row by row through a cyclic shift operation. The compressed sampling rate is dynamically adjusted according to the local variance of the sub-block to generate compressed observations.

[0016] For each sub-block's compressed observations, signal reconstruction is performed based on the Toeplitz cyclic measurement matrix and an improved orthogonal matching pursuit algorithm. Noise interference is suppressed by dynamically adjusting the regularization parameter, and sub-block reconstruction signals are generated.

[0017] All sub-block reconstruction signals are weighted and fused according to spatial location. Residual compensation is performed on the sub-block reconstruction signals of overlapping areas, and the reconstruction results are iteratively corrected to output high-resolution signals.

[0018] As a preferred embodiment of the DC charging pile detection method applied in the field according to the present invention, the specific steps of calculating the local variation coefficient through sliding window analysis, marking abnormal locations, and integrating them into an abnormal location set are as follows.

[0019] The sliding window of the high-resolution signal is defined by the spatial location dimension and the temporal slice dimension. The local variation coefficients of each sliding window in the temporal slice dimension and the spatial location dimension are calculated respectively. Spatiotemporal coupling is performed according to the preset local variance fusion weight to generate a spatiotemporal coupling variation coefficient distribution map.

[0020] Anomaly detection thresholds are dynamically generated based on the spatiotemporal coupling coefficient of variation distribution map, and the center point of the sliding window that exceeds the anomaly detection threshold is marked as a candidate anomaly point.

[0021] Morphological closing operations are performed on the marked candidate outliers to eliminate spatially isolated points and fill temporally continuous outlier regions, generating a denoised outlier map.

[0022] The abnormal points in the denoised abnormal map are mapped to a spatiotemporal coordinate system, and adjacent abnormal points are merged using a spatiotemporal proximity density clustering algorithm to output a set of abnormal locations representing the abnormal locations of the cable.

[0023] As a preferred embodiment of the DC charging pile testing method applied in the field according to the present invention, the specific steps for calculating the maximum allowable output current of the DC charging pile in real time are as follows:

[0024] Extract the spatial distribution parameters, time duration parameters, maximum strain gradient and average temperature gradient of each cable anomaly location from the set of anomaly locations to generate a spatiotemporal feature vector sequence.

[0025] The system uses a pre-trained long short-term memory prediction model to perform time series prediction on the spatiotemporal feature vector sequence, outputs the predicted value of the remaining insulation lifetime, and generates the fault risk level based on the mapping relationship between the predicted value of the remaining insulation lifetime and the preset resistance decay rate threshold.

[0026] The predicted remaining insulation life and fault risk level are used as state inputs. The strategy is iteratively optimized through a reinforcement learning network to output the maximum allowable output current adjustment. This adjustment is then linearly superimposed onto the current reference current value to output the maximum allowable output current of the DC charging pile.

[0027] As a preferred embodiment of the DC charging pile testing method applied in the field according to the present invention, the specific steps for extracting harmonic components are as follows:

[0028] The deviation between the maximum allowable output current of the DC charging pile and the real-time output current is calculated, and the dynamic target impedance value is calculated using Ohm's law. The impedance matching control signal is generated by decomposing it into the duty cycle control parameters in the memristor array through the pulse coding algorithm.

[0029] The total resistance of the memristor array is adjusted based on the impedance matching control signal. The output current of the DC charging pile is collected in real time, and synchronous compressed wavelet transform analysis is performed to extract the harmonic ridges with concentrated time-frequency energy. The harmonic ridges are then segmented by K-means energy density clustering to obtain the harmonic components.

[0030] As a preferred embodiment of the DC charging pile detection method applied in the field according to the present invention, the specific steps for generating the optimal network reconstruction scheme are as follows:

[0031] The spatial density distribution of each cable abnormality location in the abnormal location set is calculated, the fault risk level is normalized and coded, and the harmonic components are weighted and summed according to the harmonic order. The components are then fused through the preset health assessment fusion weights to generate the power distribution network health index.

[0032] The power distribution network health index is used as a risk quantification term, which together with minimizing network losses constitutes the optimization objective. Combined with topological constraints and voltage safety constraints, a quantum annealing optimization problem is constructed.

[0033] The quantum annealing algorithm is used to solve the quantum annealing optimization problem that satisfies topological and voltage safety constraints, and the optimal network reconstruction scheme is generated.

[0034] Secondly, the present invention provides a DC charging pile testing system for on-site application, including a data acquisition module for acquiring the strain gradient and temperature gradient of the DC charging pile output cable, and performing preprocessing to generate a three-dimensional dataset.

[0035] The anomaly detection module is used to input the 3D dataset into the Toeplitz cyclic measurement matrix for compressed sampling, and reconstruct it into a high-resolution signal using an improved orthogonal matching pursuit algorithm. It then calculates the local coefficient of variation through sliding window analysis, marks the anomaly locations, and integrates them into an anomaly location set.

[0036] The health prediction module is used to input the set of abnormal locations into a pre-trained long short-term memory prediction model to assess the remaining insulation life and fault risk level, and to calculate the maximum allowable output current of the DC charging pile in real time.

[0037] The harmonic analysis module is used to dynamically adjust the detection load impedance through a memristor array, collect the DC charging pile output current in real time, and use synchronous compressed wavelet transform to analyze the harmonic ridges and extract the harmonic components.

[0038] The scheme generation module integrates the abnormal location set, fault risk level and harmonic components, outputs the power distribution network health index, and generates the optimal network reconfiguration scheme through the quantum annealing algorithm.

[0039] Thirdly, the present invention provides a computer device including a memory and a processor, wherein the memory stores a computer program, wherein when the computer program is executed by the processor, it implements any step of the DC charging pile detection method applied in the field as described in the first aspect of the present invention.

[0040] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the DC charging pile detection method applied in the field as described in the first aspect of the present invention.

[0041] The beneficial effects of this invention are as follows: High-precision real-time monitoring of the insulation status of DC charging piles is achieved through compressed sensing reconstruction of the three-dimensional strain-temperature gradient field and memristor dynamic impedance adjustment technology. The use of Toeplitz compressed sampling and an improved orthogonal matched pursuit algorithm significantly reduces the amount of data while greatly improving the sensitivity of micro-strain detection. Combined with fast impedance matching of the memristor array and synchronous compressed wavelet transform, high-precision harmonic feature extraction is achieved. Finally, an intelligent closed-loop system is formed through LSTM lifetime prediction and quantum annealing optimization, significantly improving the accuracy of fault early warning and enabling rapid dynamic adjustment of the output current, effectively solving the technical challenge of multi-physics coupling monitoring of fast charging piles. Attached Figure Description

[0042] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0043] Figure 1 is a flowchart of the DC charging pile testing method applied in the field.

[0044] Figure 2 is a schematic diagram of a DC charging pile testing system applied in the field.

[0045] Figure 3 is a flowchart of the compressed sampling and signal reconstruction process.

[0046] Figure 4 is a flowchart of the power distribution network health index generation and network reconfiguration process. Detailed Implementation

[0047] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0048] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0049] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0050] Referring to Figures 1 to 4, an embodiment of the present invention is provided, which offers a method for testing DC charging piles in the field, comprising the following steps:

[0051] S1. Collect the strain gradient and temperature gradient of the DC charging pile output cable, and perform preprocessing to generate a three-dimensional dataset.

[0052] GPS timestamps were used to align the strain and temperature gradients in time, and adaptive Kalman filtering and median filtering were used to eliminate noise interference. The filtered strain and temperature gradients were then interpolated to a unified spatiotemporal grid according to spatial location, time slice, and physical quantity type to generate a three-dimensional dataset.

[0053] It should be noted that the raw data of strain gradient (example sampling rate 1kHz) and temperature gradient (example sampling rate 100Hz) are synchronously recorded by triggering the data acquisition card through the PPS signal, and the time axis of the two is unified to the GPS second pulse clock reference, and the time alignment error is controlled within 1ms.

[0054] For strain gradients, adaptive Kalman filtering is used. The process noise covariance matrix is ​​dynamically updated based on the local variance calculated by the sliding window (example window length 50). The measurement noise covariance matrix is ​​fixed to the sensor calibration value. For temperature gradients, 5×5 neighborhood median filtering is used to eliminate impulse noise. The filtered strain gradients and temperature gradients are filled with missing values ​​and resampled to a unified grid point through cubic spline interpolation according to three dimensions: spatial coordinates (example interval 1cm), time slices (example interval 10ms), and physical quantity type (strain gradient or temperature gradient). Finally, a three-dimensional dataset containing spatial location, timestamp, and physical quantity value is generated.

[0055] It should also be noted that the measurement noise covariance matrix is ​​directly adopted from the sensor's factory calibration value (for example, the measurement noise variance calibration of a strain sensor is 1×10). -4 με 2, where με represents micro-strain), the sensor's factory calibration value remains fixed during the filtering process; the process noise covariance matrix is ​​obtained by calculating the local statistical characteristics of the strain gradient signal in real time.

[0056] S2. The three-dimensional dataset is input into the Toeplitz cyclic measurement matrix for compressed sampling, and reconstructed into a high-resolution signal using an improved orthogonal matching pursuit algorithm. The local variation coefficient is calculated through sliding window analysis, anomaly locations are marked, and integrated into an anomaly location set.

[0057] The 3D dataset is divided into multiple sub-blocks based on spatial location, time slice, and physical quantity type.

[0058] It should be noted that the spatial location dimension is divided into fixed sizes (e.g., a rectangular grid of 10cm × 10cm), the time slice dimension is divided into fixed durations (e.g., 100ms as a time slice unit), and the physical quantity type dimension keeps the strain gradient and temperature gradient independently divided.

[0059] For the generated 3D dataset containing spatial coordinates, timestamps, and physical quantity values, 10cm × 10cm spatial regions are sequentially extracted along the xy-plane coordinate axes. Within each spatial region, all 3D data points within a corresponding 100ms time window are extracted and categorized into strain gradient sub-blocks and temperature gradient sub-blocks. A sliding window method with a 20% overlap rate (2cm spatial overlap and 20ms temporal overlap in the example) is used to ensure boundary continuity. Each sub-block data is stored as a five-dimensional tensor structure (x-coordinate, y-coordinate, time point, strain gradient value, and temperature gradient value), with missing data points filled using nearest-neighbor interpolation. The final output is a set of sub-blocks named according to spatial grid number, time slice number, and physical quantity type.

[0060] The first row of elements is generated based on a random Gaussian distribution for each sub-block, and a Toeplitz cyclic measurement matrix is ​​generated row by row through a cyclic shift operation. The compressed sampling rate is dynamically adjusted according to the local variance of the sub-block to generate compressed observations.

[0061] It should be noted that the three-dimensional data (spatial location, time slice, physical quantity type) within the sub-block is unfolded into a one-dimensional vector. A sliding window method is used to move and calculate along the vector. The window size is fixed (e.g., a window of 500 three-dimensional data points composed of 5×5 spatial units and 20 time slices), sliding gradually with a 50% overlap. At each window position, the Welford online algorithm is used to calculate the unbiased variance of the sub-block sliding window (example: 5×5 spatial units × 20 time slices) in real time. After traversing the entire sub-block vector, the arithmetic mean of the variance values ​​calculated for all windows is taken as the final local variance of that sub-block.

[0062] The first row of elements is randomly generated using a Gaussian distribution with a mean of zero. The number of elements is equal to the sub-block dimension d (e.g., when the sub-block is expanded into a 1000-dimensional vector, 1000 independent and identically distributed random numbers of N(0,1 / d) are generated). A Toeplitz cyclic measurement matrix is ​​constructed by a cyclic right shift operation. Specifically, the first row of elements is shifted right by k-1 positions to generate the k-th row of elements (k ranges from 1 to the number of rows m in the measurement matrix), forming an m×d dimensional Toeplitz cyclic measurement matrix. The compressed sampling rate is dynamically adjusted. Finally, the Toeplitz cyclic measurement matrix is ​​multiplied by the vectorized sub-block data to output compressed observations.

[0063] It should also be noted that the base compression sampling rate is set according to the relationship between measurement dimension and signal sparsity in compressed sensing theory, and the compression sampling rate is dynamically adjusted according to the local variance of the sub-blocks. The expression is as follows:

[0064] η = 1 - 0.5log2(1 + σ) 2 / σ 2 );

[0065] Where η is the compressed sampling rate (ranging from 0 to 1), σ 2 It is the local variance of the sub-block, σ 2 It is the preset noise variance;

[0066] Preset noise variance is usually determined in advance based on hardware parameters or empirical data and is used for noise calibration in algorithms.

[0067] For each sub-block's compressed observations, signal reconstruction is performed based on the Toeplitz cyclic measurement matrix and an improved orthogonal matching pursuit algorithm. Noise interference is suppressed by dynamically adjusting the regularization parameter, thus generating the sub-block reconstructed signal.

[0068] It should be noted that the Toeplitz cyclic measurement matrix is ​​combined with an improved orthogonal matching pursuit algorithm to progressively reconstruct the signal by iteratively selecting atoms and calculating residuals. In each iteration, the improved orthogonal matching pursuit algorithm calculates the inner product of the current residual and each column of the Toeplitz cyclic measurement matrix, selects the column with the largest absolute value of the inner product as the new atom of the support set, updates the signal estimate using the least squares method, and recalculates the residual. The regularization parameter is dynamically adjusted according to the ratio of the sub-block local variance to the preset noise variance. For example, when the sub-block local variance is less than the preset noise variance, the regularization parameter is increased to enhance noise suppression. The iteration termination condition is set to the residual norm reaching a preset maximum number of iterations (e.g., 100 times), and the final output is the sub-block reconstructed signal.

[0069] All sub-block reconstruction signals are weighted and fused according to spatial location. Residual compensation is performed on the sub-block reconstruction signals of overlapping areas, and the reconstruction results are iteratively corrected to output high-resolution signals.

[0070] It should be noted that a spatial weight mapping table is established based on the spatial coordinates of the sub-block reconstructed signals. The weight value is inversely proportional to the distance from the sub-block center point to the fusion point (the distance from the sub-block center point to the fusion point is calculated using existing three-dimensional Euclidean distance methods). For multiple sub-block reconstructed signals in the overlapping region, a weighted average algorithm is used for preliminary fusion, with the weight coefficients determined by the spatial weight mapping. For the fusion boundary region, the residual vectors of adjacent sub-block reconstructed signals are calculated, and the residual compensation coefficients are solved using the least squares method to compensate and correct the sub-block reconstructed signals in the overlapping region. An iterative approach is used to optimize the fusion result. In each iteration, the residual compensation coefficients are updated and the weighted fusion value is recalculated until the mean square error of adjacent sub-block reconstructed signals in the overlapping region reaches the maximum number of iterations (e.g., 10), resulting in a high-resolution output signal.

[0071] It should also be noted that the spatial coordinates (e.g., three-dimensional coordinates (x, y, z)) of the center point of the reconstructed signal of each sub-block are obtained, the weights of the target fusion point and the center points of each sub-block are calculated using the Gaussian kernel function, the weights of all relevant sub-blocks are normalized, and finally a spatial weight mapping table containing normalized weights and corresponding coordinate indices is generated.

[0072] A sliding window for the high-resolution signal is defined according to the spatial location dimension and the temporal slice dimension. The local variation coefficients of each sliding window in the temporal slice dimension and spatial location dimension are calculated respectively. Spatiotemporal coupling is performed according to the preset local variance fusion weight to generate a spatiotemporal coupling variation coefficient distribution map.

[0073] It should be noted that a rectangular window (e.g., 5×5 cells) is set in the spatial location dimension, and a fixed-length window (e.g., 20 slices) is set in the time slice dimension. The window slides on the 3D data volume with a 50% overlap rate. At each window position, the local coefficient of variation (time slice standard deviation divided by time slice mean) in the time slice dimension and the local coefficient of variation (spatial location standard deviation divided by spatial location mean) in the spatial location dimension are calculated respectively.

[0074] The local coefficients of variation in the two dimensions are weighted and summed according to preset local variance fusion weights (e.g., 0.6 for the time dimension and 0.4 for the spatial dimension) to obtain the spatiotemporal coupling coefficient of variation. After traversing the complete data volume, the spatiotemporal coupling coefficients of variation at all window positions are arranged according to spatial coordinates and timestamps to generate a spatiotemporal coupling coefficient of variation distribution map.

[0075] It should be noted that the preset local variance fusion weights are determined by analyzing the signal stability of the time slice and spatial location dimensions, assigning lower weights to dimensions with larger fluctuations (e.g., when the time dimension fluctuates significantly, the time weight is set to 0.3-0.5); secondly, referring to sensor accuracy parameters, the weights of dimensions with higher measurement noise are correspondingly reduced (e.g., the variance of strain gradient spatial measurement noise is calibrated to 1×10). -4με 2 When the spatial weight is increased to 0.6-0.7, the weight combination that optimizes the signal-to-noise ratio of the spatiotemporal coupling coefficient of variation distribution map is selected through experiments.

[0076] Anomaly detection thresholds are dynamically generated based on the spatiotemporal coupling coefficient of variation distribution map, and the center point of the sliding window that exceeds the anomaly detection threshold is marked as a candidate anomaly point.

[0077] It should be noted that the global mean and global standard deviation of the spatiotemporal coupling coefficient of variation distribution map are calculated using the arithmetic mean and standard deviation. The global mean plus three times the global standard deviation is used as the initial anomaly judgment threshold. A three-dimensional sliding window statistical analysis is performed on the spatiotemporal coupling coefficient of variation distribution map. The window size is set to three time slices multiplied by three spatial location units multiplied by three spatial location units. The local mean and local standard deviation of the spatiotemporal coupling coefficient of variation within each window range are calculated. When the coefficient value of the center point of the window (the spatiotemporal coupling coefficient of variation value corresponding to the center position within the spatiotemporal region covered by the sliding window) exceeds the local mean plus twice the local standard deviation, the center point is marked as a candidate anomaly point.

[0078] Morphological closing operations are performed on the marked candidate outliers to eliminate spatially isolated points and fill temporally continuous outlier regions, generating a denoised outlier map.

[0079] It should be noted that the occurrence of candidate outliers at each spatial location is checked in the time slice dimension. When the same spatial coordinate is marked as a candidate outlier in multiple consecutive time slices (e.g., 3 slices), it is determined that there is a temporally continuous outlier in this region, and it is marked as a temporally continuous outlier region. A circular structuring element (e.g., radius 5 cm) is defined in the spatial location dimension, and an expansion operation is performed on all candidate outliers to merge isolated points whose spatial distance is less than the diameter of the structuring element into connected regions. Subsequently, an erosion operation is performed to restore the original boundary of the outlier region but eliminate unconnected isolated points. A linear structuring element (e.g., length 3 consecutive slices) is defined in the time slice dimension, and expansion and erosion operations are performed on the outlier sequence at each spatial location to fill in the anomalous time segments with brief interruptions. The final generated outlier map is denoised.

[0080] The abnormal points in the denoised abnormal map are mapped to a spatiotemporal coordinate system, and adjacent abnormal points are merged using a spatiotemporal proximity density clustering algorithm to output a set of abnormal locations representing the abnormal locations of the cable.

[0081] It should be noted that the process involves reading the denoised anomaly map after morphological closing operations, where pixel values ​​of 1 represent potential anomalies; scanning the entire 3D data structure of the anomaly map, checking the marking status of each spatial location on each time slice, and recording the coordinates of all pixels marked as 1; extracting the complete spatiotemporal coordinate information of these anomalies, including their spatial location (x, y, z) and corresponding time slice number; and finally matching and verifying this coordinate information with the original cable location mapping table to obtain the anomalies in the denoised anomaly map.

[0082] Based on the spatial coordinates and time slice information of each anomaly point in the denoised anomaly map, the precise location of the anomaly points is reconstructed in a three-dimensional spatiotemporal coordinate system. A density-based spatiotemporal clustering algorithm is used, setting the spatial neighborhood radius (e.g., 10 cm) and temporal neighborhood range (e.g., 5 consecutive slices) as clustering parameters. Adjacent anomalies that satisfy the density connectivity condition are merged. During the clustering process, Euclidean distance is used to calculate the anomaly proximity in the spatial dimension, and the number of slice intervals is used to calculate continuity in the temporal dimension. When anomalies satisfy the neighborhood condition in both spatial distance and time interval, they are merged into the same anomaly cluster. The final output set of anomaly locations includes the center coordinates, time span, and spatial coverage information of each anomaly cluster. The spatial coverage is determined by the minimum bounding rectangle of all anomalies within the cluster, and the time span is determined by the earliest and latest time slices of the anomalies within the cluster.

[0083] S3. Input the set of abnormal locations into the pre-trained long short-term memory prediction model to assess the remaining insulation life and fault risk level, and calculate the maximum allowable output current of the DC charging pile in real time.

[0084] Extract the spatial distribution parameters, time duration parameters, maximum strain gradient, and average temperature gradient of each abnormal location from the set of abnormal locations to generate a spatiotemporal feature vector sequence.

[0085] It should be noted that, for each cable anomaly location in the set of anomaly locations, the spatial distribution parameters are calculated, including the coordinates of the center of the anomaly region and the length, width, and height of the spatial coverage area (the coordinates of the center of the anomaly region and the length, width, and height of the spatial coverage area are calculated using the minimum bounding rectangle algorithm); the statistical time duration parameters include the start time slice, the end time slice, and the total number of duration slices; the strain gradient data of the anomaly location on all time slices within the duration are obtained by querying the original 3D dataset, and the maximum strain gradient value is extracted; the average temperature gradient value of the spatial region corresponding to the anomaly location within the duration is calculated.

[0086] The features of four dimensions—spatial distribution parameters, time duration parameters, maximum strain gradient, and average temperature gradient—are arranged in a fixed order (spatial distribution parameters (center coordinates, length, width, and height), time duration parameters (start and end slices, duration), maximum strain gradient, and average temperature gradient) to form a spatiotemporal feature vector characterizing the location of a single cable anomaly. This process is repeated for all anomaly locations, arranging the spatiotemporal feature vectors of each anomaly location in chronological order, ultimately forming a complete sequence of spatiotemporal feature vectors.

[0087] By using a pre-trained long short-term memory prediction model, time series prediction is performed on the spatiotemporal feature vector sequence, outputting the predicted value of the remaining insulation lifetime, and generating the fault risk level based on the mapping relationship between the predicted value of the remaining insulation lifetime and the preset resistance decay rate threshold.

[0088] It should be noted that the spatiotemporal feature vector sequence is input into the Long Short-Term Memory (LSTM) prediction model in chronological order. The feature vectors are processed step-by-step through memory units and gating mechanisms to output a sequence of predicted remaining insulation lifetime values. Subsequently, a mapping table between the predicted remaining insulation lifetime values ​​and the resistance decay rate is established (e.g., a predicted remaining insulation lifetime of 2000 hours corresponds to a resistance decay rate of 5%, and 1000 hours corresponds to 10%). Risk levels are classified according to preset resistance decay rate thresholds (e.g., resistance decay rate <8% is low risk, 8%-15% is medium risk, and >15% is high risk). Finally, an assessment report containing the predicted remaining insulation lifetime values ​​and corresponding fault risk levels is generated.

[0089] It should also be noted that the training process of the long short-term memory prediction model uses historical cable monitoring data, the input is a spatiotemporal feature vector sequence, the label is the measured remaining insulation life value, the prediction error is minimized by the Adam optimizer, the number of training iterations is set to 200 rounds, the batch size is set to 32, and the optimal long short-term memory prediction model parameters are determined by 5-fold cross-validation.

[0090] The predicted remaining insulation life and fault risk level are used as state inputs. The strategy is iteratively optimized through a reinforcement learning network to output the maximum allowable output current adjustment. This adjustment is then linearly superimposed onto the current reference current value to output the maximum allowable output current of the DC charging pile.

[0091] It should be noted that a two-dimensional state vector is formed by combining the predicted remaining insulation life and the fault risk level, and then input into a pre-trained reinforcement learning network. The reinforcement learning network outputs a continuous action value, i.e., the maximum allowable output current adjustment, with the adjustment range limited to ±50A of the example value. The rated output current value (e.g., 250A) specified on the charging pile nameplate is used as the basic reference value. The output current adjustment is algebraically added to the current reference current value, expressed as:

[0092] I max =Ibase +ΔI;

[0093] Among them, I max This is the maximum allowable output current of a DC charging pile (unit: amperes), I base ΔI is the current reference current value (unit: ampere), and ΔI is the current adjustment amount output by the deep deterministic policy gradient network (unit: ampere), ΔI∈[-50,+50];

[0094] It should also be noted that the state space of the reinforcement learning network consists of the predicted remaining insulation lifetime and the fault risk level, the action space is the current adjustment, and the reward function is designed as a weighted sum of the cable safety factor and charging efficiency, with weight coefficients of 0.7 and 0.3 respectively (example values). The reinforcement learning network training employs an experience replay mechanism, sampling state-action-reward samples from historical operating data for batch training, and optimizing the network parameters through a policy gradient algorithm. During training, a target reinforcement learning network and a soft update strategy are used to ensure stability, with a discount factor of 0.9 (example value) and a policy network learning rate of 0.001 (example value). Each training round undergoes 10,000 iterations.

[0095] S4. The load impedance is dynamically adjusted and detected by the memristor array, the output current of the DC charging pile is collected in real time, and the harmonic ridge is analyzed and the harmonic components are extracted by synchronous compressed wavelet transform.

[0096] The deviation between the maximum allowable output current of the DC charging pile and the real-time output current is calculated, and the dynamic target impedance value is calculated using Ohm's law. The impedance matching control signal is generated by decomposing the dynamic target impedance value into the duty cycle control parameters in the memristor array through the pulse coding algorithm.

[0097] It should be noted that the real-time output current is obtained by sampling in real time using a high-precision Hall current sensor connected in series at the output terminal of the DC charging pile. The deviation from the maximum allowable output current of the DC charging pile is calculated using the following expression:

[0098] δI=∣I max -I real |;

[0099] Where δI is the current deviation (unit: A), I max This is the maximum allowable output current of a DC charging pile (unit: amperes), I real It is the real-time output current (unit: A);

[0100] The dynamic target impedance value can be calculated using Ohm's law, expressed as:

[0101]

[0102] Among them, Ztarget It is the dynamic target impedance value, V out It is the output voltage of the DC charging pile, which is measured in real time by a voltage sensor (e.g., with an accuracy of ±0.2%).

[0103] A mapping table between dynamic target impedance values ​​and duty cycle parameters is established. This table, obtained through experimental calibration, contains 256 discrete levels (e.g., an impedance value of 10Ω corresponds to a duty cycle of 50%). The dynamic target impedance value is quantized using a pulse code algorithm. Based on the mapping table, the closest discrete level is matched, and the corresponding duty cycle parameter is output. The duty cycle parameter is converted into a pulse width modulation signal with a frequency set to 1kHz. The pulse width is proportional to the duty cycle parameter. Finally, an impedance matching control signal is generated.

[0104] The total resistance of the memristor array is adjusted based on the impedance matching control signal. The output current of the DC charging pile is collected in real time, and synchronous compressed wavelet transform analysis is performed to extract the harmonic ridges with concentrated time-frequency energy. The harmonic ridges are then segmented by K-means energy density clustering to obtain the harmonic components.

[0105] It should be noted that the real-time resistance of each memristor cell in the memristor array is measured at a set control voltage (e.g., 1V) (example values ​​are 1kΩ-100kΩ), the admittance values ​​(reciprocals of the resistance values) of all parallel cells are added together, and the reciprocal is taken to obtain the total resistance of the memristor array.

[0106] The total resistance of the memristor array is adjusted according to the impedance matching control signal to match the output circuit impedance of the DC charging pile with the dynamic target impedance value. The adjusted output current signal is collected synchronously with a sampling frequency of 10kHz. A five-level synchronous compressed wavelet transform is performed using the db4 wavelet basis function to obtain the time-frequency energy distribution matrix. Continuous time-frequency points with energy exceeding the energy threshold (e.g., twice the overall energy average) are extracted from the time-frequency energy distribution matrix to form harmonic ridges. K-means clustering analysis is performed on the extracted harmonic ridges, with the cluster number K=3 (corresponding to the fundamental, 3rd harmonic, and 5th harmonic). Clustering is performed using the energy value of the time-frequency points as a feature, and finally, the time-frequency distribution region of each harmonic component and its energy proportion are output.

[0107] It should also be noted that the output current signal of the DC charging pile under normal operating conditions is collected, and the time-frequency energy distribution data is extracted by synchronous compressed wavelet transform. The global mean and standard deviation of the energy at all time-frequency points are calculated, and the energy threshold is set to the global mean plus twice the standard deviation (for example, when the global mean is 50dB, the energy threshold is set to 70dB).

[0108] S5 integrates the abnormal location set, fault risk level, and harmonic components to output the power distribution network health index, and generates the optimal network reconfiguration scheme through the quantum annealing algorithm.

[0109] The spatial density distribution of each cable abnormality location in the abnormal location set is calculated, the fault risk level is normalized and coded, and the harmonic components are weighted and summed according to the harmonic order. The components are then fused using preset health assessment fusion weights to generate a power distribution network health index.

[0110] It should be noted that the straight-line distance between the center points of adjacent cables (e.g., 2 meters between cables A and B) is directly measured from the power distribution network topology drawing and used as the physical spacing d. The kernel function bandwidth is set to d / 2 (example value: h = 1 meter when d = 2 meters). A Gaussian kernel function is applied to each location representing a cable anomaly. By superimposing the kernel function contributions of all anomaly locations, the spatial density distribution value representing the cable anomaly location is calculated using a kernel density estimation algorithm. The expression is:

[0111]

[0112] in, It represents the spatial density distribution value at the coordinates (x, y) of the cable anomaly location, n represents the total number of cable anomaly locations, and h is the kernel function bandwidth (unit: meters). i ,y i (x,y) is the two-dimensional spatial coordinate of the i-th cable anomaly location, i is the index variable representing the cable anomaly location, K is the Gaussian kernel function, and (x,y) is the coordinate of the cable anomaly location.

[0113] The coordinates of the cable anomaly locations are extracted from the denoised anomaly map using a three-dimensional spatiotemporal clustering algorithm (DBSCAN variant), specifically the centroid coordinates (spatial x, y, z) and time slice range of each anomaly cluster.

[0114] The fault risk levels are normalized and coded according to the International Electrotechnical Commission (IEC) insulation assessment standards. High-risk levels correspond to a significant decrease in insulation resistance, coded with higher values; medium-risk levels correspond to a moderate decrease in insulation resistance, coded with intermediate values; and low-risk levels correspond to a slight decrease in insulation resistance, coded with lower values. Next, harmonic components are frequency-domain weighted, with the fundamental component assigned the highest weight, the third harmonic component assigned a medium weight, and the fifth harmonic component assigned a smaller weight. This weighting allocation follows the IEEE harmonic standards' provisions regarding the effects of each harmonic. Finally, a linear weighted fusion is performed using health assessment fusion weighting coefficients determined by expert consensus. The cable abnormal spatial density distribution value is assigned the highest weight reflecting the degree of defect aggregation, the risk level code value is assigned a medium weight reflecting the degree of insulation degradation, and the harmonic weighted sum is assigned a smaller weight reflecting the impact of current distortion. The standard range of the distribution network health index is obtained through weighted summation.

[0115] The power distribution network health index is used as a risk quantification term, which together with minimizing network losses constitutes the optimization objective. Combining topological constraints and voltage safety constraints, a quantum annealing optimization problem is constructed.

[0116] It should be noted that the distribution network health index is used as a risk quantification term, which, together with minimizing network losses, constitutes a multi-objective optimization function. A linear weighting method is used to transform the dual objectives into a single objective, with the weight coefficients determined based on the degree of risk preference. Secondly, topological constraints are established, including node connectivity constraints, branch capacity constraints, and radial operation constraints. Simultaneously, voltage safety constraints are set, requiring that the voltage deviation of each node does not exceed the allowable fluctuation range of the rated voltage. The above objective function and constraints are mapped to the Hamiltonian of a quantum annealing problem, where the objective function is converted into an energy term, and hard constraints are achieved through penalty terms. The final result is a quantum annealing optimization problem.

[0117] It should also be noted that topology constraints are achieved by establishing a node-branch association matrix, which requires three conditions to be met: network connectivity constraints, ensuring that all load nodes and power supply nodes remain connected; branch capacity constraints, ensuring that the power of each line does not exceed its maximum transmission capacity; and radial operation constraints, which use a depth-first search algorithm to verify that the network is acyclic.

[0118] Voltage safety constraints are achieved by setting upper and lower limits for node voltages, requiring the voltage amplitude of each node to be maintained within a specified percentage range of the rated voltage (e.g., ±5%). Power flow calculations are performed using the Newton-Raphson method for verification.

[0119] The quantum annealing algorithm is used to solve the quantum annealing optimization problem that satisfies topological and voltage safety constraints, and the optimal network reconstruction scheme is generated.

[0120] It should be noted that the power distribution network reconfiguration problem is mapped to a quantum annealing model, where the network switching states are represented by qubits, with a closed state encoded as 1 and an open state encoded as 0. Next, a Hamiltonian containing the objective function and constraints is constructed. The objective function consists of minimizing network losses and optimizing the health of the power distribution network. The constraints include topological connectivity constraints, branch capacity constraints, and voltage safety constraints. An annealing operation is performed using a quantum annealing processor, with an initial temperature set to a high value, gradually decreasing the temperature until the qubit states converge to the optimal solution. Finally, the qubit states are decoded to obtain the switching operation scheme, generating the optimal network reconfiguration scheme that satisfies all constraints.

[0121] This embodiment also provides a DC charging pile testing system for on-site application, including: a data acquisition module for acquiring the strain gradient and temperature gradient of the DC charging pile output cable, and performing preprocessing to generate a three-dimensional dataset;

[0122] The anomaly detection module is used to input the 3D dataset into the Toeplitz cyclic measurement matrix for compressed sampling, and reconstruct it into a high-resolution signal using an improved orthogonal matching pursuit algorithm. It then calculates the local coefficient of variation through sliding window analysis, marks the anomaly locations, and integrates them into an anomaly location set.

[0123] The health prediction module is used to input the set of abnormal locations into a pre-trained long short-term memory prediction model to assess the remaining insulation life and fault risk level, and to calculate the maximum allowable output current of the DC charging pile in real time.

[0124] The harmonic analysis module is used to dynamically adjust the detection load impedance through a memristor array, collect the DC charging pile output current in real time, and use synchronous compressed wavelet transform to analyze the harmonic ridges and extract the harmonic components.

[0125] The scheme generation module integrates the abnormal location set, fault risk level and harmonic components, outputs the power distribution network health index, and generates the optimal network reconfiguration scheme through the quantum annealing algorithm.

[0126] This embodiment also provides a computer device applicable to the on-site DC charging pile testing method, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the on-site DC charging pile testing method proposed in the above embodiment.

[0127] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0128] This embodiment also provides a storage medium storing a computer program, which, when executed by a processor, implements the DC charging pile detection method proposed in the above embodiments for application in the field. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0129] In summary, this invention achieves high-precision real-time monitoring of the insulation status of DC charging piles through compressed sensing reconstruction of the three-dimensional strain-temperature gradient field and memristor dynamic impedance adjustment technology. The use of Toeplitz compressed sampling and an improved orthogonal matched pursuit algorithm significantly reduces the amount of data while greatly improving the sensitivity of micro-strain detection. Combined with fast impedance matching of the memristor array and synchronous compressed wavelet transform, high-precision harmonic feature extraction is achieved. Finally, an intelligent closed-loop system is formed through LSTM lifetime prediction and quantum annealing optimization, significantly improving the accuracy of fault warning and enabling rapid dynamic adjustment of the output current, effectively solving the technical challenge of multi-physics coupling monitoring of fast charging piles.

[0130] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for testing DC charging piles in the field, characterized in that: This includes collecting the strain gradient and temperature gradient of the DC charging pile output cable, preprocessing them, and generating a three-dimensional dataset. Specifically, the strain gradient and temperature gradient are time-aligned using GPS timestamps, and adaptive Kalman filtering and median filtering are used to eliminate noise interference. The filtered strain gradient and temperature gradient are then interpolated to a unified spatiotemporal grid according to spatial location, time slice, and physical quantity type to generate a three-dimensional dataset. The three-dimensional dataset is input into the Toeplitz cyclic measurement matrix for compressed sampling, and reconstructed into a high-resolution signal using an improved orthogonal matching pursuit algorithm. Local variation coefficients are calculated using sliding window analysis to mark abnormal locations, which are then integrated into an abnormal location set. This abnormal location set is input into a pre-trained long short-term memory prediction model to assess the remaining insulation life and fault risk level, and to calculate the maximum allowable output current of the DC charging pile in real time. The load impedance is dynamically adjusted using a memristor array, and the output current of the DC charging pile is acquired in real time. Synchronous compressed wavelet transform is used to analyze harmonic ridges and extract harmonic components. The abnormal location set, fault risk level, and harmonic components are fused to output a distribution network health index. The optimal network reconstruction scheme is then generated using a quantum annealing algorithm. The specific steps are as follows: the spatial density distribution of each cable abnormal location in the abnormal location set is calculated; the fault risk level is normalized and encoded; the harmonic components are weighted and summed according to the harmonic order; and the results are fused using preset health assessment fusion weights to generate the distribution network health index. The power distribution network health index is used as a risk quantification term, which together with minimizing network loss constitutes the optimization objective. Combining topological constraints and voltage safety constraints, a quantum annealing optimization problem is constructed. The quantum annealing algorithm is used to solve the quantum annealing optimization problem that satisfies the topological constraints and voltage safety constraints, generating the optimal network reconfiguration scheme.

2. The DC charging pile testing method applied in the field as described in claim 1, characterized in that: The specific steps for inputting the 3D dataset into the Toeplitz cyclic measurement matrix for compressed sampling and reconstructing it into a high-resolution signal using an improved orthogonal matching pursuit algorithm are as follows: The 3D dataset is divided into multiple sub-blocks according to spatial location, time slice, and physical quantity type; the first row of elements is generated based on each sub-block using a random Gaussian distribution, and the Toeplitz cyclic measurement matrix is ​​generated row by row through a cyclic shift operation; the compressed sampling rate is dynamically adjusted according to the local variance of the sub-blocks to generate compressed observations. For each sub-block's compressed observations, signal reconstruction is performed based on the Toeplitz cyclic measurement matrix and an improved orthogonal matching pursuit algorithm. Noise interference is suppressed by dynamically adjusting the regularization parameter, and sub-block reconstruction signals are generated. All sub-block reconstruction signals are weighted and fused according to spatial location. Residual compensation is performed on the sub-block reconstruction signals of overlapping areas, and the reconstruction results are iteratively corrected to output high-resolution signals.

3. The DC charging pile testing method applied in the field as described in claim 1, characterized in that: The process of calculating local variation coefficients through sliding window analysis, marking abnormal locations, and integrating them into an abnormal location set involves the following steps: A sliding window for the high-resolution signal is defined according to the spatial location dimension and the time slice dimension. The local variation coefficients for each sliding window in both the time slice dimension and the spatial location dimension are calculated. Spatiotemporal coupling is performed according to a preset local variance fusion weight to generate a spatiotemporal coupling variation coefficient distribution map. An anomaly judgment threshold is dynamically generated based on the spatiotemporal coupling variation coefficient distribution map. The center points of the sliding windows exceeding the anomaly judgment threshold are marked as candidate anomaly points. Morphological closing operations are performed on the marked candidate anomaly points to eliminate spatially isolated points and fill temporally continuous anomaly regions, generating a denoised anomaly map. The anomaly points in the denoised anomaly map are mapped to a spatiotemporal coordinate system, and adjacent anomaly points are merged using a spatiotemporal proximity density clustering algorithm, outputting an abnormal location set representing the cable anomaly location.

4. The DC charging pile testing method applied in the field as described in claim 1, characterized in that: The specific steps for calculating the maximum allowable output current of the DC charging pile in real time are as follows: extract the spatial distribution parameters, time duration parameters, maximum strain gradient and average temperature gradient of each abnormal location from the abnormal location set to generate a spatiotemporal feature vector sequence; perform time series prediction on the spatiotemporal feature vector sequence through a pre-trained long short-term memory prediction model to output the insulation remaining life prediction value, and generate the fault risk level based on the mapping relationship between the insulation remaining life prediction value and the preset resistance attenuation rate threshold. The predicted remaining insulation life and fault risk level are used as state inputs. The strategy is iteratively optimized through a reinforcement learning network to output the maximum allowable output current adjustment. This adjustment is then linearly superimposed onto the current reference current value to output the maximum allowable output current of the DC charging pile.

5. The DC charging pile testing method applied in the field as described in claim 1, characterized in that: The specific steps for extracting harmonic components are as follows: The deviation between the maximum allowable output current of the DC charging pile and the real-time output current is calculated, and the dynamic target impedance value is calculated using Ohm's law. This is then decomposed into duty cycle control parameters in the memristor array using a pulse coding algorithm to generate an impedance matching control signal. Based on the impedance matching control signal, the total resistance of the memristor array is adjusted. The output current of the DC charging pile is collected in real time, and synchronous compressed wavelet transform analysis is performed to extract harmonic ridges with concentrated time-frequency energy. K-means energy density clustering is then performed on the harmonic ridges to obtain the harmonic components.

6. A DC charging pile testing system for on-site application, based on the DC charging pile testing method for on-site application according to any one of claims 1 to 5, characterized in that: This includes a data acquisition module, used to collect the strain gradient and temperature gradient of the DC charging pile output cable, and perform preprocessing to generate a three-dimensional dataset; The anomaly detection module is used to input the 3D dataset into the Toeplitz cyclic measurement matrix for compressed sampling, and reconstruct it into a high-resolution signal using an improved orthogonal matching pursuit algorithm. It then calculates the local coefficient of variation through sliding window analysis, marks the anomaly locations, and integrates them into an anomaly location set. The health prediction module is used to input the set of abnormal locations into a pre-trained long short-term memory prediction model to assess the remaining insulation life and fault risk level, and to calculate the maximum allowable output current of the DC charging pile in real time. The harmonic analysis module is used to dynamically adjust the load impedance through a memristor array, collect the output current of the DC charging pile in real time, and use synchronous compressed wavelet transform to analyze the harmonic ridges and extract harmonic components. The scheme generation module integrates the abnormal location set, fault risk level and harmonic components, outputs the power distribution network health index, and generates the optimal network reconstruction scheme through quantum annealing algorithm.

7. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the DC charging pile testing method applied in the field as described in any one of claims 1 to 5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the DC charging pile testing method applied to the field as described in any one of claims 1 to 5.

Images