Method for identifying abnormal power consumption behavior of charging pile based on deep learning

Abstract

The invention relates to the technical field of intelligent power systems and deep learning, in particular to a charging pile abnormal power consumption behavior identification method based on deep learning. The method comprises the following steps: introducing dynamic time correlation modeling, and analyzing power consumption time sequence data to obtain time dynamic characteristics; based on the power utilization time sequence data, capturing the spatial relationship and mutual influence between different charging piles to obtain spatial correlation characteristics; fusing the time dynamic feature and the space correlation feature to generate a spatio-temporal feature after multi-scale fusion; and carrying out nonlinear mapping and classification based on the spatio-temporal characteristics after multi-scale fusion, and judging an anomaly type. The problems that a traditional charging pile abnormal power consumption behavior recognition method mainly depends on simple rule judgment, complex space-time dependency cannot be processed, and the phenomenon of false alarm or missing alarm is likely to occur are solved; the problems that space correlation between the charging piles is neglected mostly, cooperative information of the multiple charging piles cannot be effectively utilized for anomaly detection, and the anomaly detection precision is low are solved.

Description

Technical Field

[0001] The present invention relates to the fields of intelligent power systems and deep learning technologies, and in particular to a method for identifying abnormal power usage behavior of charging piles based on deep learning. Background Art

[0002] With the increasing popularity of electric vehicles, charging piles, as crucial infrastructure for charging electric vehicles, have been widely deployed. The stability and safety of charging piles are directly related to the proper use of electric vehicles and the safety of users. However, during actual operation, charging piles may experience abnormal power usage, such as current overloads, voltage instability, and power surges, due to equipment failures, improper user operation, or changes in environmental factors. These abnormalities not only affect charging efficiency but also pose safety risks, such as battery damage, equipment damage, and even fire risks.

[0003] Therefore, timely and accurate detection and identification of abnormal power usage in charging piles, ensuring the safety of the charging process and the reliability of the equipment, has become a critical task in current charging pile management. With the increasing number of charging piles and the continuous advancement of technology, improving the accuracy and efficiency of anomaly detection through more intelligent methods has become a pressing technical challenge. Therefore, the use of deep learning technology to intelligently identify abnormal power usage in charging piles has important practical significance and application value.

[0004] Traditional methods for identifying abnormal power usage behavior in charging piles have the following technical problems: they mainly rely on simple rule judgments or threshold-based monitoring methods, and are unable to handle complex spatiotemporal dependencies, resulting in poor sensitivity to small fluctuations or sudden anomalies, and are prone to false alarms or missed alarms; most of them ignore the spatial correlation between charging piles, and are unable to effectively utilize the collaborative information of multiple charging piles for anomaly detection. They have difficulty coping with the complex behavior patterns of charging piles under different environmental conditions, and have low anomaly detection accuracy. Summary of the Invention

[0005] The present invention provides a method for identifying abnormal power usage behavior of charging piles based on deep learning, so as to solve the problems that traditional methods for identifying abnormal power usage behavior of charging piles mainly rely on simple rule judgment or threshold-based monitoring methods, cannot handle complex spatiotemporal dependencies, resulting in poor sensitivity to small fluctuations or sudden anomalies, and are prone to false alarms or missed alarms; most of them ignore the spatial correlation between charging piles, cannot effectively utilize the collaborative information of multiple charging piles for anomaly detection, have difficulty in dealing with the complex behavior patterns of charging piles under different environmental conditions, and have low anomaly detection accuracy.

[0006] The present invention provides a method for identifying abnormal electricity usage behavior of charging piles based on deep learning, which specifically includes the following technical solutions: A method for identifying abnormal power usage behavior of charging piles based on deep learning, comprising the following steps: S1. Obtain the electricity consumption data of charging piles and record it in the form of time series to generate electricity consumption time series data; introduce dynamic time correlation modeling to analyze the electricity consumption time series data and obtain temporal dynamic characteristics; based on the electricity consumption time series data, introduce spatial correlation modeling to capture the spatial relationship and mutual influence between different charging piles and obtain spatial correlation characteristics; S2. Through the spatiotemporal scale transformation model, the temporal dynamic features and spatial correlation features are fused to generate multi-scale fused spatiotemporal features. Based on the multi-scale fused spatiotemporal features, nonlinear mapping and classification are performed to obtain a multi-dimensional probability vector. Based on the multi-dimensional probability vector, the anomaly type is determined and abnormal power usage behavior of charging piles is identified.

[0007] Preferably, the S1 specifically includes: In the process of dynamic time correlation modeling, the electricity consumption time series data is transformed and convolved to extract time series features.

[0008] Preferably, the S1 specifically includes: Based on the time series features, the time dynamic features are calculated in the form of weighted sum and combined with the time decay term.

[0009] Preferably, the S1 specifically includes: In the process of spatial correlation modeling, spatial features are extracted from electricity consumption time series data through spatial convolution operations.

[0010] Preferably, the S1 specifically includes: Based on spatial characteristics, the spatial weight coefficient is introduced, and the spatial correlation characteristics are calculated by combining the spatiotemporal joint weight term.

[0011] Preferably, the S2 specifically includes: The spatiotemporal scale transformation model introduces different scale angles, introduces a transformation matrix at each scale, performs weighted fusion of temporal dynamic features and spatial correlation features, and obtains spatiotemporal features.

[0012] Preferably, the S2 specifically includes: The spatiotemporal features of all scales are linearly transformed and then weightedly fused to obtain multi-scale fused spatiotemporal features.

[0013] Preferably, the S2 specifically includes: The spatiotemporal features after multi-scale fusion are used as the input of the multi-layer perceptron model, and nonlinear mapping and discrimination are performed to obtain a multidimensional probability vector. The maximum value of the multidimensional probability vector is determined to identify abnormal behavior of the charging pile.

[0014] The beneficial effects of the technical solution of the present invention are: 1. Through multi-level and multi-scale deep learning processing such as dynamic time correlation modeling, spatial correlation modeling, and spatiotemporal scale transformation modeling, the present invention can accurately capture various small fluctuations and sudden anomalies in the power consumption process of charging piles, and identify various abnormal behaviors of charging piles in complex environments, such as overload, current instability, and power surge.

[0015] 2. Through the fusion of spatiotemporal features and multi-scale modeling, it is possible to fully utilize electricity consumption time series data, spatial information, and spatiotemporal dependencies at different scales, thereby improving the accuracy and robustness of anomaly recognition. Utilizing deep learning models, it automatically learns and optimizes the underlying patterns in charging pile electricity consumption data, reducing the need for manual intervention. Adaptive adjustments are made based on actual data, continuously optimizing recognition performance to adapt to changes in different charging piles and usage environments.

[0016] 3. The deep learning method of the present invention is not only applicable to charging piles of different models and brands, but can also adapt to the abnormal behavior identification needs in different scenarios according to the ever-changing power demand and external environment; through continuous training of the deep learning model, it can adapt to more types of abnormal behavior patterns and has good scalability. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is a flow chart of a method for identifying abnormal electricity usage behavior of charging piles based on deep learning according to the present invention. DETAILED DESCRIPTION

[0018] In order to further illustrate the technical means and effects adopted by the present invention to achieve the predetermined purpose of the invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0019] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0020] The following describes in detail a method for identifying abnormal electricity usage behavior of charging piles based on deep learning provided by the present invention with reference to the accompanying drawings.

[0021] Refer to the attached Figure 1 , which shows a flow chart of a method for identifying abnormal electricity usage behavior of a charging pile based on deep learning provided by one embodiment of the present invention. The method includes the following steps: S1. Obtain the electricity consumption data of charging piles and record it in the form of time series to generate electricity consumption time series data; introduce dynamic time correlation modeling to analyze the electricity consumption time series data and obtain temporal dynamic characteristics; based on the electricity consumption time series data, introduce spatial correlation modeling to capture the spatial relationship and mutual influence between different charging piles and obtain spatial correlation characteristics; Smart meters or sensors integrated into charging piles collect electrical parameters such as voltage, current, and power in real time, capturing the charging pile's electricity usage data. Specifically, each charging pile is equipped with power monitoring equipment that records electrical parameters such as voltage, current, and power at a frequency of seconds or even milliseconds during the charging process. The frequency and granularity of data collection can be adjusted based on actual needs to ensure that even minor fluctuations and sudden anomalies during the charging process are captured.

[0022] The electricity consumption data of the charging pile (electrical parameters such as current, voltage and power) are continuously recorded in the form of time series to obtain electricity consumption time series data. In order to fully explore the dependency relationship between different time points in the electricity consumption time series data, dynamic time association modeling is introduced. Based on the long short-term memory network, the transform domain convolutional network is used to perform transform convolution operations on the electricity consumption time series data, output time series features, and generate time dynamic features based on the time series dependency relationship. Specifically, the electricity consumption time series data of the charging pile is set to ,in Representative time point Power consumption data (such as current, voltage or power values) on is the length of the time series; time points The dynamic characteristics of need to comprehensively consider the correlation between the current moment and the past moment. Therefore, the temporal dependency is expressed in the form of weighted sum combined with the attenuation function. The temporal dynamic characteristics formula of each moment is:

[0023] in, Indicates the current time point The temporal dynamic characteristics of the time reflect the dynamic relationship between the current moment and the past moment; is the length of the time series, including all time points from the past to the present; Indicates a historical point in time For the current time point the intensity of the impact; is the time decay term, which is used to quantify the timeliness of historical influence; It is a historical point in time The associated weight coefficient is optimized by gradient descent method and its value range is [0,1]; It is an index of a historical time point; is the current time point; It is a historical point in time The time decay coefficient is used to control the degree of decay of the dependency between time points. It is initialized using the statistical characteristics of the charging pile's power consumption time series data and optimized using the gradient descent method. The value range is [0.01, 0.1]. The statistical characteristics refer to the analysis results of factors such as the change trend, periodicity, lag effect, and abnormal response duration in the historical power consumption time series data. They are used to assist in determining the dependency strength and decay rate between time points. Statistical methods are well known to those skilled in the art and will not be described in detail here. It is through the transformation convolution operation from the historical time point Electricity usage data of charging piles The time series features extracted from the data are normalized during feature extraction, so there is no dimension problem. The normalization method is a well-known technical means for those skilled in the art and will not be described in detail here. is the convolution kernel weight; It's time The bias term is used to correct the time series features and serve as the transform domain convolution network parameters and convolution kernel weights. The training method of the transform domain convolutional network obtained by synchronous training is a technical means well known to those skilled in the art and will not be described in detail here; the gradient descent method is a technical means well known to those skilled in the art and will not be described in detail here.

[0024] Since charging piles are distributed in different spatial locations, there may be correlation between the electricity usage behaviors of charging piles. For example, adjacent charging piles may be affected by the same external environmental factors and show similar electricity usage patterns. Therefore, based on convolutional neural networks, spatial correlation modeling is introduced to capture the spatial relationship and mutual influence between different charging piles. By considering the spatial dependence between charging piles, the interaction between different charging piles can be understood and abnormal electricity usage behaviors can be effectively identified. is the location coordinate set of the charging pile, where is the number of charging piles, Is the location index of the charging station, ranging from 1 to , the spatial correlation characteristics of each charging pile are calculated using the following formula:

[0025] in, Indicates the The spatial correlation characteristics of the charging piles reflect the spatial relationship between the current charging pile and other charging piles and the correlation of power behavior caused by the spatial location; For the The first charging station and the The spatial weight coefficient between charging piles has a value range of [0,1] and is calculated using a distance metric (such as Euclidean distance). The calculation method is well known to those skilled in the art and will not be described in detail here. is the spatiotemporal joint weight term, used to quantify the Charging piles at historical time points The electricity consumption data of Charging piles at the current time The dynamic influence intensity of spatial correlation characteristics; is the time weight coefficient, which means that Charging pile and When considering the spatial relationship of the charging piles, the importance of each time point is obtained through the joint optimization of attention mechanism and gradient descent, and the value range is [0,1]; is the attenuation coefficient of spatial correlation, reflecting the The charging pile is The influence degree of each charging pile is optimized by gradient descent method, and the value range is [0.001, 5]. To perform spatial convolution operations at historical time points From Electricity consumption data of charging piles The spatial features extracted from is the convolution kernel weight, which is obtained through training optimization; For the The spatial bias term of each charging pile is used to adjust the spatial correlation characteristics of each charging pile, which is initialized to a zero vector and obtained through training optimization. The attention mechanism and gradient descent method, i.e., the model training optimization method, are technical means well known to those skilled in the art and will not be described in detail here. Spatial relationships are used to enhance the spatial correlation model's ability to learn similar power usage behaviors between charging piles, helping to identify abnormal power usage behaviors caused by similar spatial layouts.

[0026] S2. Through the spatiotemporal scale transformation model, the temporal dynamic features and spatial correlation features are fused to generate multi-scale fused spatiotemporal features. Based on the multi-scale fused spatiotemporal features, nonlinear mapping and classification are performed to obtain a multi-dimensional probability vector. Based on the multi-dimensional probability vector, the anomaly type is determined and abnormal power usage behavior of charging piles is identified.

[0027] The electricity consumption behavior of charging piles may show different patterns at different time scales and spatial scales. In order to better extract spatiotemporal features and fuse information at different scales, a spatiotemporal scale transformation model is introduced to fuse spatiotemporal features from different scale perspectives to enhance the ability to identify abnormal behavior. Specifically, multiple scales are introduced. At each scale, the spatial correlation features and temporal dynamic features are weighted through the corresponding transformation matrix to obtain spatiotemporal features. The spatial and temporal information is integrated to ensure that the spatiotemporal features at different scales can fully reflect the abnormal behavior of the charging piles, so as to capture multi-level spatiotemporal relationships and thus enhance the ability to identify abnormal behavior. The spatiotemporal features at each scale are calculated using the following formula:

[0028] in, For the The spatiotemporal characteristics at different scales reflect the spatiotemporal correlation characteristics of charging piles at different scales; and Respectively The transformation matrix of spatial correlation features and temporal dynamic features at different scales is used to control the fusion of features at different scales, and is obtained by the gradient descent method. The gradient descent method is a technical means well known to those skilled in the art and will not be described in detail here; For the The scale bias term of each scale is used to adjust the spatiotemporal characteristics at each scale. The gradient is calculated by the back propagation algorithm and updated using an optimizer (such as SGD and Adam). The back propagation algorithm and optimizer update method are technical means well known to those skilled in the art and will not be described in detail here. The spatiotemporal characteristics of the charging pile are captured from different scale angles through the spatiotemporal scale transformation model.

[0029] The spatiotemporal features of all scales of the output are subjected to linear transformation and weighted fusion, and a classification bias term is added to obtain the multi-scale fused spatiotemporal features; the multi-scale fused spatiotemporal features are used as the input of the multi-layer perceptron (MLP) model, and then processed through multiple fully connected layers and activation functions for nonlinear mapping and discrimination to obtain the final classification output; the multi-layer perceptron (MLP) model is a technical means well known to those skilled in the art and will not be described in detail here; the final classification output is a multi-dimensional probability vector representing the probability distribution of abnormal states, and each dimension in the multi-dimensional probability vector represents the prediction confidence of the multi-layer perceptron model for a certain abnormal type, such as overload, power surge, charging interruption, etc. The number of dimensions is consistent with the number of abnormal types. The calculation formula of the multi-dimensional probability vector is as follows:

[0030] in, It is a multidimensional probability vector that represents the probability distribution of each possible abnormal power consumption behavior. It is used to identify whether there is abnormal behavior in the charging pile and its abnormality type (such as overload, short circuit, etc.); Indicates the The weight matrix of each scale is responsible for mapping the spatiotemporal features to the final classification space. It is initialized using Xavier and automatically learned through the back propagation algorithm. The Xavier method and the back propagation algorithm are both well-known technical means to those skilled in the art and will not be described in detail here. It is a classification bias term, which is used to translate the spatiotemporal features after linear weighted summation, so that the multi-layer perceptron model has stronger representation ability. Its value range is [-1,1]; Assume there is abnormal type (for example: overload, short circuit, current instability, etc.), then the multidimensional probability vector is a -dimensional vector, where: Represents the probability of "overload" occurring; Represents the probability of "short circuit"; and so on, Represents the probability of the last abnormal behavior; By multidimensional probability vector The maximum value of is used to determine the current abnormal behavior of the charging pile. For example, if If it is the maximum value and it represents "current instability", it is judged that the current charging pile has abnormal behavior of unstable current, thus providing a basis for subsequent management and maintenance.

[0031] In summary, a method for identifying abnormal power consumption behavior of charging piles based on deep learning has been completed.

[0032] The order in which the embodiments of the invention are presented is for illustrative purposes only and does not necessarily represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0033] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments.

[0034] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included in the scope of protection of the present invention.

Claims

1. A method for identifying abnormal electricity usage behavior of charging piles based on deep learning, characterized in that: The following steps are involved: S1. Obtain the electricity consumption data of the charging pile and record it in the form of time series to generate electricity consumption time series data; introduce dynamic time correlation modeling to analyze the electricity consumption time series data and obtain time dynamic characteristics; Based on the electricity consumption time series data, spatial correlation modeling is introduced to capture the spatial relationship and mutual influence between different charging piles and obtain spatial correlation characteristics; S2. Through the spatiotemporal scale transformation model, the temporal dynamic features and spatial correlation features are fused to generate multi-scale fused spatiotemporal features. Based on the multi-scale fused spatiotemporal features, nonlinear mapping and classification are performed to obtain a multi-dimensional probability vector. Based on the multi-dimensional probability vector, the anomaly type is determined and abnormal power usage behavior of charging piles is identified.

2. The method for identifying abnormal electricity usage behavior of charging piles based on deep learning according to claim 1 is characterized in that: Said S1 specifically includes: In the process of dynamic time correlation modeling, the electricity consumption time series data is transformed and convolved to extract time series features.

3. The method for identifying abnormal electricity usage behavior of charging piles based on deep learning according to claim 2 is characterized in that: Said S1 specifically includes: Based on the time series features, the time dynamic features are calculated in the form of weighted sum and combined with the time decay term.

4. The method for identifying abnormal electricity usage behavior of charging piles based on deep learning according to claim 3 is characterized in that: Said S1 specifically includes: In the process of spatial correlation modeling, spatial features are extracted from electricity consumption time series data through spatial convolution operations.

5. The method for identifying abnormal electricity usage behavior of charging piles based on deep learning according to claim 4 is characterized in that: Said S1 specifically includes: Based on spatial characteristics, the spatial weight coefficient is introduced, and the spatial correlation characteristics are calculated by combining the spatiotemporal joint weight term.

6. The method for identifying abnormal electricity usage behavior of charging piles based on deep learning according to claim 1 is characterized in that: Said S2 specifically includes: The spatiotemporal scale transformation model introduces different scale angles, introduces a transformation matrix at each scale, performs weighted fusion of temporal dynamic features and spatial correlation features, and obtains spatiotemporal features.

7. The method for identifying abnormal electricity usage behavior of charging piles based on deep learning according to claim 6 is characterized in that: Said S2 specifically includes: The spatiotemporal features of all scales are linearly transformed and then weightedly fused to obtain multi-scale fused spatiotemporal features.

8. The method for identifying abnormal electricity usage behavior of charging piles based on deep learning according to claim 7 is characterized in that: Said S2 specifically includes: The spatiotemporal features after multi-scale fusion are used as the input of the multi-layer perceptron model, and nonlinear mapping and discrimination are performed to obtain a multidimensional probability vector. The maximum value of the multidimensional probability vector is determined to identify abnormal behavior of the charging pile.

Images