Charging pile with electronic leakage protection technology and method thereof

By obtaining and analyzing the residual current signal in the charging circuit in the charging pile, identifying the risk of electronic leakage and automatically cutting off the power supply, the problem of electronic leakage risk during charging of electric vehicles is solved, the safety of charging piles is improved, and electric shocks and fires are prevented. The accident occurred.

CN116639007BActive Publication Date: 2024-11-12JIANGXI DONKEY CHARGING TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202310765077.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-06-27
Publication Date
2024-11-12
Estimated Expiration
2043-06-27

AI Technical Summary

Technical Problem

The risk of electronic leakage caused by charging equipment failure or aging of wires during electric vehicles may cause electric shock accidents or fires, and it is difficult for existing technology to effectively prevent it.

Method used

A charging pile with electronic leakage protection technology is designed. By obtaining the residual current signal in the charging circuit, multi-scale correlation feature vector analysis is performed, and the charging power is automatically cut off based on the analysis results to prevent safety accidents caused by electronic leakage.

Benefits of technology

By monitoring the current situation in the charging circuit in real time, accurately identifying the risk of electronic leakage, cutting off the charging power supply in a timely manner, improving the safety of charging piles, avoiding electric shocks and fire accidents, and providing a safer and more reliable charging experience.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN116639007B_ABST
    Figure CN116639007B_ABST
Patent Text Reader

Abstract

The present invention discloses a charging pile with electronic leakage protection technology and a method thereof, which obtains a residual current signal in a charging circuit; analyzes the residual current signal to obtain a residual current multi-scale correlation feature vector; and determines whether to automatically cut off the charging power supply based on the residual current multi-scale correlation feature vector. In this way, it is possible to detect whether there is an electronic leakage risk based on the residual current signal in the charging circuit, thereby determining whether to automatically cut off the charging power supply, thereby improving the safety of the charging pile and avoiding fire accidents caused by electronic leakage of the charging pile.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of intelligent charging technology, and in particular to a charging pile with electronic leakage protection technology and a method thereof. Background Art

[0002] With the popularity of electric vehicles, the demand for charging piles is also increasing. The function of charging piles is similar to that of gas pumps in gas stations. They can be fixed on the ground or walls and installed in public buildings such as public buildings, shopping malls, public parking lots, and residential parking lots or charging stations. They can charge various types of electric vehicles according to different voltage levels. The input end of the charging pile is directly connected to the AC power grid, and the output end is equipped with a charging plug for charging electric vehicles. At present, during the charging process of electric vehicles, there may be a risk of electronic leakage in the charging pile due to failure of the charging equipment, aging or damage of the wires, etc., posing a threat to the safety of users. If measures are not taken in time, electronic leakage may cause electric shock accidents or fires.

[0003] Therefore, an optimized charging pile with electronic leakage protection technology is desired. Summary of the invention

[0004] The embodiment of the present invention provides a charging pile with electronic leakage protection technology and a method thereof, which obtains a residual current signal in a charging circuit; analyzes the residual current signal to obtain a residual current multi-scale correlation feature vector; and determines whether to automatically cut off the charging power supply based on the residual current multi-scale correlation feature vector. In this way, it is possible to detect whether there is an electronic leakage risk based on the residual current signal in the charging circuit, thereby determining whether to automatically cut off the charging power supply, thereby improving the safety of the charging pile and avoiding fire accidents caused by electronic leakage of the charging pile.

[0005] The embodiment of the present invention further provides a charging pile with electronic leakage protection technology, which includes:

[0006] A residual current signal acquisition system is used to obtain the residual current signal in the charging circuit;

[0007] A signal analysis system, used for analyzing the residual current signal to obtain a residual current multi-scale correlation feature vector; and

[0008] The charging power supply control system is used to determine whether to automatically cut off the charging power supply based on the residual current multi-scale correlation feature vector.

[0009] The embodiment of the present invention further provides a charging method with electronic leakage protection technology, which includes:

[0010] Obtaining the residual current signal in the charging circuit;

[0011] Analyzing the residual current signal to obtain a residual current multi-scale correlation feature vector; and

[0012] Based on the residual current multi-scale correlation feature vector, it is determined whether to automatically cut off the charging power supply. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. In the drawings:

[0014] Figure 1 A block diagram of a charging pile with electronic leakage protection technology provided in an embodiment of the present invention.

[0015] Figure 2 A block diagram of a signal analysis system in a charging pile with electronic leakage protection technology provided in an embodiment of the present invention.

[0016] Figure 3 A block diagram of a training module in a charging pile with electronic leakage protection technology provided in an embodiment of the present invention.

[0017] Figure 4 The present invention provides a flow chart of a charging method with electronic leakage protection technology in an embodiment of the present invention.

[0018] Figure 5 A schematic diagram of a system architecture of a charging method with electronic leakage protection technology provided in an embodiment of the present invention.

[0019] Figure 6 This is an application scenario diagram of a charging pile with electronic leakage protection technology provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0020] To make the purpose, technical solution and advantages of the embodiments of the present invention more clear, the embodiments of the present invention are further described in detail below in conjunction with the accompanying drawings. Here, the exemplary embodiments of the present invention and their descriptions are used to explain the present invention, but are not intended to limit the present invention.

[0021] Unless otherwise specified, all technical and scientific terms used in the embodiments of the present application have the same meaning as those commonly understood by those skilled in the art of the present application. The terms used in this application are only for the purpose of describing specific embodiments and are not intended to limit the scope of this application.

[0022] In the embodiments of the present application, it should be noted that, unless otherwise specified and limited, the term "connection" should be understood in a broad sense. For example, it can be an electrical connection or a connection between two components. It can be a direct connection or an indirect connection through an intermediate medium. For ordinary technicians in this field, the specific meanings of the above terms can be understood according to the specific circumstances.

[0023] It should be noted that the terms "first\second\third" involved in the embodiments of the present application are only used to distinguish similar objects, and do not represent a specific order for the objects. It is understandable that the specific order or sequence of "first\second\third" can be interchanged where permitted. It should be understood that the objects distinguished by "first\second\third" can be interchanged where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein.

[0024] At present, during the charging process of electric vehicles, there may be a risk of electronic leakage in the charging pile due to the failure of the charging equipment, aging or damage of the wires, etc., which poses a threat to the safety of users. If measures are not taken in time, electronic leakage may cause electric shock accidents or fires. Therefore, an optimized charging pile with electronic leakage protection technology is desired.

[0025] In one embodiment of the present invention, Figure 1 FIG. 1 is a block diagram of a charging pile with electronic leakage protection technology provided in an embodiment of the present invention. Figure 1 As shown, a charging pile 100 with electronic leakage protection technology according to an embodiment of the present invention includes: a residual current signal acquisition system 110, used to obtain a residual current signal in a charging circuit; a signal analysis system 120, used to analyze the residual current signal to obtain a residual current multi-scale correlation feature vector; and a charging power supply control system 130, used to determine whether to automatically cut off the charging power supply based on the residual current multi-scale correlation feature vector.

[0026] The residual current signal acquisition system 110 can accurately acquire the residual current signal in the charging circuit. To ensure accuracy, the acquisition system has high sensitivity and high sampling rate to capture weak residual current signals. By acquiring the residual current signal, the current situation in the charging circuit can be monitored in real time, providing a data basis for subsequent signal analysis.

[0027] The signal analysis system 120 analyzes the collected residual current signal to extract the multi-scale correlation feature vector of the residual current, which includes signal filtering, time-frequency analysis, feature extraction and other processing steps. By analyzing the residual current signal, the risk of electronic leakage in the charging circuit can be identified, and the multi-scale correlation feature vector can provide more comprehensive and accurate information, which is helpful to determine whether the charging power supply needs to be cut off.

[0028] The charging power supply control system 130 determines whether there is an electronic leakage risk based on the residual current multi-scale correlation feature vector obtained by signal analysis, and automatically cuts off the charging power supply when necessary. The charging power supply control system has a fast response and reliable cut-off mechanism to ensure the safety of the charging process. By automatically cutting off the charging power supply, electric shock accidents and fires caused by electronic leakage can be effectively avoided. This improves the safety of the charging pile and protects the safety of users and equipment.

[0029] Charging piles with electronic leakage protection technology monitor and protect the safety of the charging process by collecting, analyzing and controlling residual current signals. They can cut off the charging power supply in time, reduce the risk of electric shock and fire, and provide users with a safer and more reliable charging experience.

[0030] Specifically, the residual current signal acquisition system 110 is used to obtain the residual current signal in the charging circuit. Accordingly, considering that in the actual operation of the charging pile, the risk of electronic leakage of the charging pile may cause a fire accident. The charging pile with electronic leakage protection technology is a device that can effectively prevent electric shock accidents and electrical fires during the charging process of electric vehicles.

[0031] It should be understood that electronic leakage protection technology is an important safety protection measure, which is widely used in charging piles, household appliances, industrial equipment and other fields. The key to electronic leakage protection technology is to accurately determine whether the residual current exceeds the set value. By collecting and analyzing the residual current signal, the risk of electronic leakage can be discovered in time, and corresponding measures can be taken to avoid electric shock and equipment damage.

[0032] In this application, electronic leakage protection technology refers to the use of electronic components and sensors to detect the residual current in the charging circuit. When the residual current exceeds the set value, the charging power supply is automatically cut off to protect the safety of personnel and equipment. Therefore, in the technical solution of this application, it is expected to detect whether there is a risk of electronic leakage based on the residual current signal in the charging circuit, so as to determine whether to automatically cut off the charging power supply, so as to improve the safety of the charging pile and avoid fire accidents caused by electronic leakage of the charging pile.

[0033] Specifically, in the technical solution of the present application, first, a residual current signal in the charging circuit is obtained. Residual current refers to the current flowing through the non-target circuit in the charging circuit, which may be caused by electronic leakage due to insulation damage of electrical equipment, line failure or other reasons. If measures are not taken in time to cut off the charging power supply, these electronic leakages may cause electric shock accidents and fires.

[0034] By collecting and analyzing the residual current signal, the multi-scale correlation feature vector of the residual current in the charging circuit can be obtained. The multi-scale correlation feature vector can reflect the degree and trend of electronic leakage, so as to determine whether there is an electronic leakage risk. If the analysis results show that the electronic leakage risk exceeds the set safety threshold, the charging power control system will automatically cut off the charging power to avoid potential dangers.

[0035] In other words, the collection and analysis of the residual current signal provides a basis for judging whether it is necessary to cut off the charging power supply. This correlation enables the charging pile to monitor the safety of the charging process in real time and take automatic protection measures when necessary to ensure the safety of users and equipment.

[0036] Specifically, the signal analysis system 120 is used to analyze the residual current signal to obtain a residual current multi-scale correlation feature vector. Figure 2 FIG. 1 is a block diagram of a signal analysis system in a charging pile with electronic leakage protection technology provided in an embodiment of the present invention, such as Figure 2 As shown, the signal analysis system 120 includes: a Gram angle and field transformation module 121, which is used to perform Gram angle and field transformation on the residual current signal to obtain a residual current timing matrix; a matrix division module 122, which is used to perform matrix division on the residual current timing matrix to obtain multiple residual current local matrices; and a multi-scale correlation analysis module 123, which is used to perform multi-scale correlation feature analysis on the multiple residual current local matrices to obtain the residual current multi-scale correlation feature vector.

[0037] It should be understood that the residual current signal can be converted into a time matrix through the Gram angle and field transformation module. The matrix partitioning module divides the time matrix into multiple local matrices, which can capture the residual current characteristics on different time scales. The multi-scale correlation analysis module analyzes these local matrices, extracts multi-scale correlation features, and more comprehensively describes the characteristics of the residual current signal.

[0038] Gram angle and field transform is an effective signal processing method that can convert time domain signals into frequency domain features. Through matrix partitioning and multi-scale correlation analysis, feature extraction can be performed at different scales, making full use of the information of residual current signals and improving the efficiency and accuracy of feature extraction.

[0039] The multi-scale correlation feature vector can more accurately reflect the characteristics of the residual current, including periodicity, volatility, abnormal peaks, etc. By analyzing these characteristics, it is possible to more reliably determine whether the residual current exceeds the set value, and take timely measures to cut off the charging power supply, thereby improving the safety of the charging pile.

[0040] That is, the combination of the Gram angle and field transformation module, the matrix partitioning module and the multi-scale correlation analysis module in the signal analysis system can effectively extract the multi-scale correlation feature vector of the residual current, improve the accuracy and efficiency of the residual current detection, and enhance the safety protection capability of the charging pile.

[0041] First, for the Gram angle and field transformation module 121, it should be understood that the Gram angle field (Gramianangular field, GAF) is based on the Gram principle. It can migrate the time series in the classical Cartesian coordinate system to the polar coordinate system for representation, which can not only retain the complete information of the signal, but also well retain the dependence and correlation of the residual current signal on time, and has similar timing characteristics to the original signal. In particular, GAF can obtain the Gram angle sum field (Gramian angular sum field, GASF) and the Gram angle difference field (Gramian angular difference field, GADF) according to the different trigonometric functions used for encoding. Since GADF is irreversible after conversion, in the technical solution of the present application, the GASF conversion method that can be reversely converted is selected to encode the residual current signal. That is to say, specifically, the residual current signal is subjected to Gram angle and field transformation to obtain the residual current timing matrix. In a specific example, the encoding steps of the residual current signal to the residual current timing matrix are as follows: for a time series with C dimensions = {Q1, Q2, ..., QC}, where each dimension contains n sampling points Qi = {qi1, qi2, ..., qin}, first normalize the data of each dimension. After that, integrate all values ​​in the data into [-1, 1]. After integration, replace the normalized values ​​with the trigonometric function value Cos value, and replace the Cartesian coordinates with polar coordinates, so as to retain the absolute time relationship of the sequence.

[0042] Furthermore, the application of Gram angle and field transformation can be used to perform time-frequency analysis on the residual current signal, thereby obtaining the time series matrix of the residual current. This transformation can convert the signal from the time domain to the time-frequency domain, providing richer information. Through Gram angle and field transformation, the residual current signal can be decomposed into time series matrices of different frequency components. These time series matrices can reflect the changes of different frequency components over time, thereby revealing the hidden features in the signal.

[0043] By analyzing the residual current time series matrix, a multi-scale correlation feature vector can be extracted to determine whether there is an electronic leakage risk. This multi-scale correlation analysis can enhance the detection capability of residual current and improve the safety of charging piles.

[0044] Therefore, the beneficial effect of applying Gram angle and field transformation to obtain the residual current timing matrix can provide a more reliable signal analysis basis for the electronic leakage protection technology of charging piles.

[0045] Then, for the matrix division module 122, then, since the residual current timing matrix contains the overall information of the residual current signal in timing, the amount of data is large and it is difficult to capture the detailed feature information in the residual current signal. Therefore, in the technical solution of the present application, the residual current timing matrix is ​​further matrix-divided to obtain a plurality of residual current local matrices. It should be understood that here, each of the residual current local matrices represents a signal within a time range in the residual current timing matrix. In this way, the local details of the timing signal can be better captured to perform more accurate electronic leakage detection and power-off protection.

[0046] Matrix partitioning of the residual current timing matrix can further extract multiple local matrices of residual currents, so as to analyze and detect the risk of electronic leakage in more detail. In one embodiment of the present application, a uniform partitioning method is adopted, and under the uniform partitioning method, the residual current timing matrix is ​​uniformly divided into multiple sub-matrices of equal size. This partitioning method is simple and intuitive, and is suitable for global analysis of the entire timing matrix. Each sub-matrix represents the residual current situation in a time period, and feature extraction and analysis can be performed independently.

[0047] In another embodiment of the present application, a feature-based partitioning method is adopted. In the feature-based partitioning method, the residual current timing matrix is ​​partitioned according to its features, and the timing matrix can be divided into multiple local matrices according to features such as the residual current amplitude, frequency, and phase. This partitioning method can extract and analyze local features of the residual current in a targeted manner according to the importance and sensitivity of different features.

[0048] In this way, on the one hand, the accuracy of feature extraction can be improved. By dividing the timing matrix into multiple local matrices, the complexity of the residual current signal can be decomposed into smaller parts, making the feature extraction more accurate. Each local matrix represents the residual current characteristics in a specific time period, and these local features can be analyzed and extracted more accurately.

[0049] On the other hand, the sensitivity of abnormality detection can be improved. The abnormality of residual current often manifests as a local abnormal pattern. By dividing it into multiple local matrices, these local abnormalities can be more easily detected. Independent abnormality detection for each local matrix can increase the sensitivity to the risk of electronic leakage and cut off the charging power supply more promptly to ensure safety.

[0050] On the other hand, it can speed up the calculation. Analyzing the entire time series matrix may result in too much calculation. After dividing the time series matrix into multiple local matrices, they can be processed and calculated in parallel, thus speeding up the calculation. This is especially important for charging pile systems with high real-time requirements.

[0051] That is, further matrix division of the residual current timing matrix to obtain multiple residual current local matrices can improve the accuracy of feature extraction, enhance the sensitivity of anomaly detection, and speed up the calculation speed, thereby enhancing the ability of residual current signal analysis and improving the safety and reliability of charging piles.

[0052] Next, the multi-scale correlation analysis module 123 includes: a residual current distribution association unit, which is used to pass the multiple residual current local matrices through a converter module including an embedding layer to obtain multiple contextual residual current local feature vectors; and a multi-scale correlation encoding unit, which is used to perform deep feature association encoding on the multiple contextual residual current local feature vectors to obtain the residual current multi-scale correlation feature vector.

[0053] Furthermore, since the residual current signal may have different local features in the charging pile, such as current components of different frequencies or current changes in different time periods, the multiple residual current local matrices are encoded through a converter module including an embedding layer to extract the global correlation feature distribution information between the local time series change features of the residual current signal in each residual current local matrix, thereby obtaining multiple contextual residual current local feature vectors.

[0054] Among them, the converter module including the embedding layer is a module for feature extraction and encoding, which can convert multiple residual current local matrices into contextual residual current local feature vectors. The module is usually composed of multiple layers, including the embedding layer.

[0055] The embedding layer is a part of the module, which is used to map the input residual current local matrix into a low-dimensional representation space. This low-dimensional representation can capture the key features of the residual current signal. The embedding layer is usually composed of a fully connected layer or a convolutional layer, which can be designed according to the specific application scenario and requirements.

[0056] Other layers of the converter module may include convolutional layers, pooling layers, normalization layers, etc. The combination of these layers can further extract and encode the residual current local matrix. The structure and parameters between these layers can be optimized and adjusted according to specific tasks and data to improve the accuracy and expression of feature extraction.

[0057] Through the converter module containing the embedding layer, the key features in each residual current local matrix can be extracted and encoded through the global correlation feature distribution information. In this way, multiple contextual residual current local feature vectors can be obtained for subsequent residual current signal analysis and anomaly detection.

[0058] Specifically, in one embodiment of the present application, in a converter module including an embedding layer, an embedding layer is designed to map each residual current local matrix to a low-dimensional feature space, and a fully connected layer or a convolutional layer can be used as the embedding layer, and appropriate parameters and activation functions can be set according to actual conditions. In the converter module, convolutional layers, pooling layers, normalization layers, etc. can be used to further extract and encode the features of the residual current local matrix. These layers can capture local timing change features through convolution operations, pooling operations, etc. By integrating multiple residual current local feature vectors, multiple contextual residual current local feature vectors can be obtained, and the multiple contextual residual current local feature vectors contain global correlation feature distribution information between each local timing change feature.

[0059] It should be understood that the converter module can learn the time series variation characteristics and global correlation feature distribution information of the residual current signal, thereby extracting more accurate features, which can better describe the characteristics of the residual current and help with subsequent analysis and processing. By extracting multiple contextual residual current local feature vectors, subtle changes and abnormalities in the residual current signal can be captured, which can enhance the ability to detect abnormal residual currents, timely discover and take corresponding measures, and improve the safety and reliability of charging piles. The residual current signal is divided into multiple local matrices and encoded through the converter module. Multiple local matrices can be processed in parallel, thereby speeding up the calculation speed, which is very important for real-time residual current signal analysis and can improve the response speed and efficiency of the charging pile.

[0060] That is, by encoding multiple residual current local matrices using a converter module including an embedding layer, the accuracy of feature extraction, the sensitivity of anomaly detection, and the speed of calculation can be improved, thereby enhancing the ability of residual current signal analysis and improving the safety and reliability of charging piles. Furthermore, the multi-scale association encoding unit is used to: arrange the multiple contextual residual current local feature vectors into one-dimensional feature vectors and then pass them through a multi-scale neighborhood feature extractor including a first convolutional layer and a second convolutional layer to obtain the residual current multi-scale association feature vector.

[0061] It is also considered that since there is an overall signal correlation relationship between the local characteristic distribution information of the residual current in each residual current local matrix, in order to fully capture and characterize the characteristics of the residual current signal, so as to accurately detect the electronic leakage situation and thus realize leakage power-off protection, in the technical solution of the present application, it is also necessary to further express the correlation characteristic information between the local characteristics of the residual current in each residual current local matrix in depth. In particular, it is also considered that in each residual current local matrix, different local areas have different contributions to the final electronic leakage detection of the correlation characteristics between the local implicit characteristic information of the residual current signal.

[0062] Therefore, the multiple contextual residual current local feature vectors are further arranged into one-dimensional feature vectors and then passed through a multi-scale neighborhood feature extractor including a first convolution layer and a second convolution layer to obtain a residual current multi-scale correlation feature vector. In particular, here, the first convolution layer and the second convolution layer respectively use one-dimensional convolution kernels of different scales to extract the multi-scale correlation feature distribution information between the various local features of the residual current signal in each residual current local matrix.

[0063] Specifically, first, each context residual current local feature vector is expanded into a one-dimensional vector, and for each context residual current local feature vector, it is expanded into a one-dimensional vector in a certain order, which can be expanded in a time sequence or other prescribed manner. Then, the expanded one-dimensional vectors are connected, and all the expanded one-dimensional vectors are connected in a certain order to form a long one-dimensional feature vector, which can be connected in the order of the context residual current local feature vectors, or adjusted according to actual needs.

[0064] Through the above steps, multiple contextual residual current local feature vectors can be arranged into a one-dimensional feature vector. The one-dimensional feature vector obtained in this way can be used as input to extract features through a multi-scale neighborhood feature extractor including a first convolutional layer and a second convolutional layer to obtain a multi-scale correlation feature vector of the residual current.

[0065] It should be understood that the multi-scale neighborhood feature extractor including the first convolution layer and the second convolution layer is a module for extracting multi-scale correlation features of residual current, which can capture features of different scales through convolution operations and extract spatial correlation in the residual current signal. Specifically, the charging power control system 130 is used to: pass the residual current multi-scale correlation feature vector through a classifier to obtain a classification result, and the classification result is used to indicate whether to automatically cut off the charging power.

[0066] Then, the residual current multi-scale associated feature vector is further passed through a classifier to obtain a classification result, and the classification result is used to indicate whether the charging power is automatically cut off. That is, in the technical solution of the present application, the label of the classifier includes automatically cutting off the charging power (first label), and not automatically cutting off the charging power (second label), wherein the classifier determines which classification label the residual current multi-scale associated feature vector belongs to through a soft maximum function.

[0067] It is worth noting that the first label p1 and the second label p2 here do not contain artificially set concepts. In fact, during the training process, the computer model does not have the concept of "whether to automatically cut off the charging power supply". It only has two classification labels and the probability of the output feature under these two classification labels, that is, the sum of p1 and p2 is one. Therefore, the classification result of whether to automatically cut off the charging power supply is actually converted from the classification label to a binary classification class probability distribution that conforms to the laws of nature. In essence, the physical meaning of the natural probability distribution of the label is used, rather than the linguistic text meaning of "whether to automatically cut off the charging power supply".

[0068] It should be understood that in the technical solution of the present application, the classification label of the classifier is a control strategy label for whether to automatically cut off the charging power supply. Therefore, after obtaining the classification result, it is possible to accurately judge whether there is electronic leakage based on the classification result, and take timely measures to cut off the charging power supply to ensure the safety of the charging pile.

[0069] The charging power control system 130 includes: a fully connected encoding unit, used to use multiple fully connected layers of the classifier to fully connect the residual current multi-scale correlation feature vector to obtain a coded classification feature vector; and a classification unit, used to pass the coded classification feature vector through the Softmax classification function of the classifier to obtain the classification result.

[0070] It should be understood that a classifier is a machine learning model that maps input data to different categories or labels. Classifiers can include: Logistic Regression, which is a widely used classification algorithm based on the linear regression model and uses a logistic function to map the output to a probability value. Logistic regression is suitable for binary classification problems. Support Vector Machine (SVM), SVM is a powerful classification algorithm that achieves classification by finding an optimal hyperplane in the feature space. SVM can handle linear and nonlinear classification problems. Decision Tree, a decision tree is a classification algorithm based on a tree structure that classifies data through a series of decision rules. Decision trees are easy to understand and interpret, but they are prone to overfitting. Random Forest, Random Forest is an ensemble learning algorithm consisting of multiple decision trees. Each decision tree classifies the data, and finally the final classification is determined by voting or averaging the prediction results. Deep learning models (such as convolutional neural networks), deep learning models perform well on complex data such as images and sequence data. Deep learning models such as Convolutional Neural Network (CNN) can be used to extract features and perform classification.

[0071] Furthermore, the charging pile with electronic leakage protection technology also includes a training module for training the converter module including the embedding layer, the multi-scale neighborhood feature extractor including the first convolution layer and the second convolution layer, and the classifier. Figure 3 A block diagram of the training module in a charging pile with electronic leakage protection technology provided in an embodiment of the present invention, such as Figure 3As shown, the training module 140 includes: a training data acquisition unit 141, used to acquire training data, the training data including a training residual current signal in a charging circuit, and a true value of whether to automatically cut off the charging power supply; a training domain conversion unit 142, used to perform Gram angle and field transformation on the training residual current signal to obtain a training residual current time series matrix; a training matrix division unit 143, used to perform matrix division on the training residual current time series matrix to obtain a plurality of training residual current local matrices; a training context-related feature extraction unit 144, used to pass the plurality of training residual current local matrices through the converter module including the embedding layer to obtain a plurality of training context residual current local feature vectors; a training multi-scale depth-related feature extraction unit 145, used to After arranging the multiple training context residual current local feature vectors into training one-dimensional feature vectors, the training residual current multi-scale associated feature vectors are obtained by passing the multi-scale neighborhood feature extractor including the first convolution layer and the second convolution layer; a classification loss unit 146 is used to pass the training residual current multi-scale associated feature vector through the classifier to obtain a classification loss function value; and a model training unit 147 is used to train the converter module including the embedding layer, the multi-scale neighborhood feature extractor including the first convolution layer and the second convolution layer, and the classifier based on the classification loss function value and through the direction propagation of gradient descent, wherein in each round of iteration of the training, the weight matrix of the classifier is iterated with the half-space structured constraint of the weight intrinsic support.

[0072] In particular, in the technical solution of the present application, after the multiple contextual residual current local feature vectors are arranged into one-dimensional feature vectors, the residual current multi-scale correlation feature vector is obtained by a multi-scale neighborhood feature extractor including a first convolutional layer and a second convolutional layer. Since each of the multiple contextual residual current local feature vectors expresses the contextual correlation local features of the matrix blocks of the residual current timing matrix, the source timing distribution differences of the respective matrix blocks of the residual current timing matrix will also cause explicit differences in the feature distributions of the respective contextual residual current local feature vectors. Therefore, when multi-scale neighborhood feature extraction is performed by a multi-scale neighborhood feature extractor including a first convolutional layer and a second convolutional layer, different feature fitting directions, such as overfitting and underfitting, will be given to the intra-vector and inter-vector local neighborhood features corresponding to each contextual residual current local feature vector.

[0073] In this way, when the obtained residual current multi-scale correlation feature vector is classified through a classifier, the partial feature distribution of the residual current multi-scale correlation feature vector corresponding to the multiple context residual current local feature vectors will also have different weight fitting directions relative to the parts corresponding to the weight matrix of the classifier. In this way, the overall feature distribution of the residual current multi-scale correlation feature vector will have a poor convergence problem relative to the weight matrix of the classifier, thereby affecting the training speed of the classifier.

[0074] Based on this, the applicant of the present application performs a half-space structured constraint of weight intrinsic support on the weight matrix M during each iteration of the weight matrix of the classifier, for example, M, during the training process of the residual current multi-scale correlation feature vector, for example, denoted as V. Specifically, in each iteration of the training, the weight matrix of the classifier is iterated with the half-space structured constraint of weight intrinsic support using the following optimization formula; wherein the optimization formula is:

[0075]

[0076] Wherein, V is the training residual current multi-scale correlation feature vector, M is the weight matrix of the classifier, V e is the matrix M T The eigenvalues ​​of M form the eigenset vector, (·) T represents the transposed vector of a vector, and denote matrix multiplication and addition respectively, and M' denotes the weight matrix of the classifier after iteration.

[0077] Here, the half-space structured constraint of the weighted eigensupport is supported by the eigenvalue set of the structured matrix of the weight matrix M of the classifier and the correlation integration of the training residual current multi-scale correlation feature vector V to be classified, to perform the structured support constraint of the hyperplane as the decision boundary on the half-space (half-space) coupled with the high-dimensional manifold of the training residual current multi-scale correlation feature vector V to be classified represented by the weight matrix M, so that the high-dimensional manifold of the training residual current multi-scale correlation feature vector V to be classified can effectively converge with respect to the hyperplane in the half-space open domain represented by the weight matrix M, thereby improving the training speed of the classifier. In this way, it is possible to accurately judge whether electronic leakage occurs based on the residual current signal in the charging circuit, and take timely measures to cut off the charging power supply when leakage is found, so as to avoid fire accidents caused by electronic leakage of the charging pile and ensure the safety of the charging pile.

[0078] In summary, a charging pile 100 with electronic leakage protection technology based on an embodiment of the present invention is explained, which detects whether there is a risk of electronic leakage based on the residual current signal in the charging circuit, and thus determines whether to automatically cut off the charging power supply, thereby improving the safety of the charging pile and avoiding fire accidents caused by electronic leakage of the charging pile.

[0079] Figure 4 The present invention provides a flow chart of a charging method with electronic leakage protection technology in an embodiment of the present invention. Figure 5 FIG. 1 is a schematic diagram of a system architecture of a charging method with electronic leakage protection technology provided in an embodiment of the present invention. Figure 4 and Figure 5 As shown, a charging method with electronic leakage protection technology includes: 210, obtaining a residual current signal in a charging circuit; 220, analyzing the residual current signal to obtain a residual current multi-scale correlation feature vector; and, 230, determining whether to automatically cut off the charging power supply based on the residual current multi-scale correlation feature vector.

[0080] Those skilled in the art will appreciate that the specific operations of each step in the above-mentioned charging method with electronic leakage protection technology have been described in the above reference. Figures 1 to 3 It has been introduced in detail in the description of the charging pile with electronic leakage protection technology, and therefore, its repeated description will be omitted.

[0081] Figure 6 FIG. 1 is an application scenario diagram of a charging pile with electronic leakage protection technology provided in an embodiment of the present invention. Figure 6 As shown, in this application scenario, first, the residual current signal in the charging circuit is obtained (for example, Figure 6 Then, the obtained residual current signal is input to a server (for example, Figure 6 In S) shown in , the server is capable of processing the residual current signal based on a charging algorithm with electronic leakage protection technology to determine whether to automatically cut off the charging power supply.

[0082] The specific embodiments described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A charging pile with electronic leakage protection technology, characterized in that: include: A residual current signal acquisition system is used to obtain the residual current signal in the charging circuit; A signal analysis system, used for analyzing the residual current signal to obtain a residual current multi-scale correlation feature vector; as well as A charging power supply control system, used to determine whether to automatically cut off the charging power supply based on the residual current multi-scale correlation feature vector; The method further includes a training module for training a converter module including an embedding layer, a multi-scale neighborhood feature extractor including a first convolutional layer and a second convolutional layer, and a classifier; Wherein, the training module includes: A training data acquisition unit, used to acquire training data, wherein the training data includes a training residual current signal in the charging circuit and a real value of whether to automatically cut off the charging power supply; A training domain conversion unit, used for performing Gram angle and field transformation on the training residual current signal to obtain a training residual current timing matrix; A training matrix division unit, used for performing matrix division on the training residual current timing matrix to obtain a plurality of training residual current local matrices; A training context associated feature extraction unit, configured to pass the plurality of training residual current local matrices through the converter module including the embedding layer to obtain a plurality of training context residual current local feature vectors; A training multi-scale deep correlation feature extraction unit, used for arranging the plurality of training context residual current local feature vectors into a training one-dimensional feature vector and then passing the plurality of training context residual current local feature vectors through the multi-scale neighborhood feature extractor including the first convolution layer and the second convolution layer to obtain a training residual current multi-scale correlation feature vector; a classification loss unit, configured to pass the training residual current multi-scale correlation feature vector through the classifier to obtain a classification loss function value; and A model training unit, configured to train the converter module including the embedding layer, the multi-scale neighborhood feature extractor including the first convolutional layer and the second convolutional layer, and the classifier based on the classification loss function value and by directional propagation of gradient descent, wherein in each round of iteration of the training, a half-space structured constraint iteration of weight intrinsic support is performed on the weight matrix of the classifier; Wherein, in each round of iteration of the training, the weight matrix of the classifier is iterated with the half-space structured constraint of the weight intrinsic support using the following optimization formula; Wherein, the optimization formula is: Wherein, V is the training residual current multi-scale correlation feature vector, M is the weight matrix of the classifier, V e is the matrix M T The eigenvalues ​​of M form the eigenvalue set vector, (·) T represents the transposed vector of a vector, and denote matrix multiplication and addition respectively, and M' denotes the weight matrix of the classifier after iteration.

2. The charging pile with electronic leakage protection technology according to claim 1 is characterized in that: The signal analysis system comprises: A Gram angle and field conversion module, used for performing Gram angle and field conversion on the residual current signal to obtain a residual current timing matrix; a matrix division module, configured to perform matrix division on the residual current timing matrix to obtain a plurality of residual current local matrices; and The multi-scale correlation analysis module is used to perform multi-scale correlation feature analysis on the multiple residual current local matrices to obtain the residual current multi-scale correlation feature vector.

3. The charging pile with electronic leakage protection technology according to claim 2 is characterized in that: The multi-scale association analysis module comprises: a residual current distribution association unit, configured to obtain a plurality of contextual residual current local feature vectors by passing the plurality of residual current local matrices through a converter module including an embedding layer; and The multi-scale correlation coding unit is used to perform deep feature correlation coding on the multiple context residual current local feature vectors to obtain the residual current multi-scale correlation feature vector.

4. The charging pile with electronic leakage protection technology according to claim 3 is characterized in that: The multi-scale association encoding unit is used to: arrange the multiple context residual current local feature vectors into a one-dimensional feature vector and then pass it through a multi-scale neighborhood feature extractor including a first convolution layer and a second convolution layer to obtain the residual current multi-scale association feature vector.

5. The charging pile with electronic leakage protection technology according to claim 4 is characterized in that: The charging power supply control system is used to: pass the residual current multi-scale correlation feature vector through a classifier to obtain a classification result, and the classification result is used to indicate whether to automatically cut off the charging power supply.

6. A charging method with electronic leakage protection technology, using the charging pile with electronic leakage protection technology according to claim 1, characterized in that: include: Obtaining the residual current signal in the charging circuit; Analyzing the residual current signal to obtain a residual current multi-scale correlation feature vector; as well as Based on the residual current multi-scale correlation feature vector, it is determined whether to automatically cut off the charging power supply.

7. The charging method with electronic leakage protection technology according to claim 6, characterized in that: Analyzing the residual current signal to obtain a residual current multi-scale correlation feature vector includes: Performing Gram angle and field transformation on the residual current signal to obtain a residual current timing matrix; Performing matrix division on the residual current timing matrix to obtain a plurality of residual current local matrices; and A multi-scale correlation feature analysis is performed on the multiple residual current local matrices to obtain the residual current multi-scale correlation feature vector.