A charging pile charging failure analysis method and system based on data mining
Through data mining and neural network technology, the problem of fault location of charging piles for new energy vehicles has been solved, rapid and accurate fault location and repair efficiency have been improved, and the accuracy of detection results has been optimized.
Patent Information
- Application Number
- CN202211680749.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-12-27
- Publication Date
- 2024-11-29
- Estimated Expiration
- 2042-12-27
AI Technical Summary
The prior art is difficult to quickly and accurately locate the charging failure of new energy vehicle charging piles, resulting in the inability to effectively identify the root cause of failure.
Using a data mining method, using DBSCAN clustering algorithm and neural network, the components and failure modes that trigger failure are determined first by obtaining the operating characteristics and failure status of the charging pile, building a charging pile fault detection model, and outputting the fault type theory Results and detection results are optimized by gradient descent model and neural network.
It realizes rapid and accurate positioning of charging pile faults, improves fault repair efficiency, and optimizes the accuracy of detection results by comparing with the sample library.
Smart Images

Figure CN115964945B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of monitoring charging piles for new energy vehicles, and more specifically to a charging pile charging failure analysis method and system based on data mining. Background Art
[0002] In recent years, my country's new energy vehicle market has shown explosive growth. By the end of 2021, the number of new energy vehicles in China reached 7.84 million, playing a huge role in alleviating resource shortages, improving environmental quality and improving energy efficiency. At present, new energy vehicles can be divided into three categories according to the power source: pure electric, hybrid and fuel cell. Among them, pure new energy vehicles have the advantages of zero emissions, high efficiency and convenient vehicle layout, thus occupying a high share in the new energy vehicle market. As we all know, the power of pure new energy vehicles mainly comes from the electric energy stored in the high-voltage battery, which can be charged by introducing external grid power through conduction. According to the charging voltage type, it can be divided into two types: AC slow charging and DC fast charging. Compared with AC slow charging, DC fast charging has higher voltage level and charging current, and its operating reliability directly affects the owner's experience and the frequency of vehicle use, and also has potential constraints on grid dispatching. Therefore, evaluating the operating reliability of DC charging piles can not only ensure the vehicle attendance rate and improve the owner's experience, but also provide a basis for formulating equipment maintenance plans, and realize the organic combination of safety, reliability and economy in the operation process of DC charging piles.
[0003] As a typical complex electronic control system, the charging pile for new energy vehicles has multiple functions such as rectification, protection, communication and output control. Its reliability during operation is affected by many factors, and the failure mechanism is complex. During the charging process, multiple faults may coexist. For example, when the cable assembly fails due to a break, the electronic lock of the charging gun also fails. Therefore, if the root cause of the failure cannot be determined, the charging fault cannot be effectively detected. However, there is no technical solution in the prior art to quickly and accurately locate the failure, making it difficult to quickly and accurately describe the charging fault of a complex electronic control system such as a charging pile.
[0004] Therefore, how to provide a charging pile charging failure analysis method and system based on data mining is a problem that technical personnel in this field urgently need to solve. Summary of the invention
[0005] In view of this, the present invention provides a charging pile charging failure analysis method and system based on data mining to solve the problems existing in the above technical solutions.
[0006] In order to achieve the above object, the present invention provides the following technical solutions:
[0007] On the one hand, the present invention provides a charging pile charging failure analysis method based on data mining, based on a charging pile for a new energy vehicle, comprising the following steps:
[0008] S100: Based on the operating characteristics of the charging pile for new energy vehicles, combined with the failure status and records of the current and previous processes, obtain its potential failure mode;
[0009] S200: Perform data mining of failure modes based on the DBSCAN clustering algorithm to obtain the number of failed components in the charging module of the charging pile for new energy vehicles;
[0010] S300: sorting the relative probabilities corresponding to different failure modes of all components in the charging module of the charging pile for the new energy vehicle according to the number of failed components, and determining the component that first triggers the failure and the triggered failure mode;
[0011] S400: Analyze the components that are determined to be the first to trigger failure and the triggered failure mode, and output theoretical results of the fault types corresponding to the charging pile for new energy vehicles.
[0012] Preferably, the potential failure modes include: performance degradation failure, functional failure and sudden failure of the charging pile for new energy vehicles.
[0013] Preferably, the S200: performing data mining of failure modes based on the DBSCAN clustering algorithm to obtain a failure checklist of components in the charging module of the charging pile for new energy vehicles, including:
[0014] S210: Count the parameters of three failure modes and standardize the parameters, including: insulation aging failure, charging interface contact wear failure, DC contactor adhesion failure, charging gun electronic lock failure, network communication failure, and cable assembly disconnection failure, to obtain a failed component data set;
[0015] S220: setting neighborhood parameters (ε, MinPts), where ε describes the neighborhood distance threshold of a component sample, and MinPts describes the threshold of the number of samples in the neighborhood of a component sample whose distance is ε; using the DBSCAN clustering algorithm, classifying the component samples of the core points on the failed component data set, and analyzing the clustering results;
[0016] S230: fixing the core points of the data of all the failed components with labels, classifying the samples of the core points on the failed component data set by using the DBSCAN clustering algorithm with adaptive feature weights, and obtaining clustering results;
[0017] S240: Count the clustering results to obtain the number of failed components. The calculation formula is as follows:
[0018] N(xi)={xj∈D|distance(xi,xj)≤Eps};
[0019] Wherein, N(xi) represents the set of failed component sample points in the Eps neighborhood of xi, and |N(xi)| is the number of samples in the failed component sample set.
[0020] Preferably, the step S300: sorting the relative probabilities corresponding to different failure modes of all components in the charging module of the charging pile for new energy vehicles according to the number of failed components, and determining the component that triggers the failure first, includes:
[0021] S310: sorting the relative probabilities corresponding to different failure modes of all components in the charging module of the charging pile for new energy vehicles;
[0022] S320: Generate a random number z1 in the interval (0, 1), determine the relative probability sorting interval position of the random number, and determine the component that first triggers the failure and the triggered failure mode.
[0023] Preferably, the S400: analyzing the components that first trigger the failure and the triggered failure mode, and outputting the theoretical results of the fault type corresponding to the charging pile for new energy vehicles:
[0024] S410: Building a charging pile fault detection model;
[0025] S420: Taking the component that first triggers failure and the triggered failure mode as input to the charging pile fault detection model, and outputting the theoretical result of the fault type corresponding to the charging pile for new energy vehicles.
[0026] Preferably, it also includes: S500: if the theoretical result of the fault type is inconsistent with the actual result of the fault type corresponding to the charging pile operating status in the pre-established sample library, then according to the pre-stored gradient descent model, the actual result of the fault type and the neural network, adjust and update the charging pile fault detection model, and re-execute step S420.
[0027] On the other hand, the present invention also provides a charging pile charging failure analysis system based on data mining, comprising:
[0028] Acquisition module: used to obtain potential failure modes based on the operating characteristics of charging piles for new energy vehicles and combined with the failure status and records of current and previous processes;
[0029] A calculation module, connected to the acquisition module, is used to perform data mining of failure modes based on a DBSCAN clustering algorithm to obtain the number of failed components in the charging module of the charging pile for new energy vehicles;
[0030] A processing module connected to the calculation module is used to sort the relative probabilities corresponding to different failure modes of all components in the charging module of the charging pile for new energy vehicles according to the number of failed components, and determine the components that trigger the failure first and the triggered failure mode;
[0031] The analysis module is connected to the processing module and is used to analyze the components that are first triggered to fail and the triggered failure mode, and output theoretical results of the fault types corresponding to the charging piles for new energy vehicles.
[0032] It can be seen from the above technical solutions that, compared with the prior art, the present invention discloses a charging pile charging failure analysis method and system based on data mining, which can efficiently and conveniently realize the failure analysis monitoring of the charging pile, and the specific beneficial effects are:
[0033] (1) The present invention first determines the component that first triggers the failure and the failure mode triggered, finds the root cause of the failure, and then obtains the corresponding fault type theoretical result. The process is simple and can quickly and accurately obtain the charging fault of a complex electronic control system such as a charging pile, thereby improving the efficiency of repairing a faulty charging pile.
[0034] (2) A neural network is used to establish charging pile fault type detection, and the theoretical results of the fault type are compared and optimized with the actual results of the fault type corresponding to the charging pile operating status in the sample library to make the detection results more accurate. BRIEF DESCRIPTION OF THE DRAWINGS
[0035] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.
[0036] Figure 1 It is a schematic diagram of the method flow of the present invention;
[0037] Figure 2 It is a schematic diagram of the system structure of the present invention. DETAILED DESCRIPTION
[0038] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions 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 creative work are within the scope of protection of the present invention.
[0039] On the one hand, see Appendix Figure 1 As shown, the embodiment of the present invention discloses a charging pile charging failure analysis method based on data mining, based on a charging pile for a new energy vehicle, comprising the following steps:
[0040] S100: Based on the operating characteristics of the charging pile for new energy vehicles, combined with the failure status and records of the current and previous processes, obtain its potential failure mode;
[0041] S200: Perform data mining of failure modes based on the DBSCAN clustering algorithm to obtain the number of failed components in the charging module of the charging pile for new energy vehicles;
[0042] S300: sorting the relative probabilities corresponding to different failure modes of all components in the charging module of the charging pile for the new energy vehicle according to the number of failed components, and determining the component that first triggers the failure and the triggered failure mode;
[0043] S400: Analyze the components that first trigger failure and the triggered failure mode, and output theoretical results of the fault type corresponding to the charging pile for new energy vehicles.
[0044] In a specific embodiment, potential failure modes include: performance degradation failure, functional failure and sudden failure of the charging pile for new energy vehicles.
[0045] In a specific embodiment, S200: performing data mining of failure modes based on the DBSCAN clustering algorithm to obtain a failure checklist of components in the charging module of the charging pile for new energy vehicles, including:
[0046] S210: Count the parameters of three failure modes and standardize the parameters, including: insulation aging failure, charging interface contact wear failure, DC contactor adhesion failure, charging gun electronic lock failure, network communication failure, and cable assembly disconnection failure, to obtain a failed component data set;
[0047] S220: setting neighborhood parameters (ε, MinPts), where ε describes the neighborhood distance threshold of a component sample, and MinPts describes the threshold of the number of samples in the neighborhood of a component sample whose distance is ε; using the DBSCAN clustering algorithm, classifying the component samples of the core points on the failed component data set, and analyzing the clustering results;
[0048] S230: fixing the core points of the data of all the failed components with labels, classifying the samples of the core points on the failed component data set by using the DBSCAN clustering algorithm with adaptive feature weights, and obtaining clustering results;
[0049] S240: Count the clustering results to obtain the number of failed components. The calculation formula is as follows:
[0050] N(xi)={xj∈D|distance(xi,xj)≤Eps};
[0051] Wherein, N(xi) represents the set of failed component sample points in the Eps neighborhood of xi, and |N(xi)| is the number of samples in the failed component sample set.
[0052] In a specific embodiment, S300: sorting the relative probabilities corresponding to different failure modes of all components in the charging module of the charging pile for new energy vehicles according to the number of failed components, and determining the component that triggers the failure first, including:
[0053] S310: sorting the relative probabilities corresponding to different failure modes of all components in the charging module of the charging pile for new energy vehicles;
[0054] S320: Generate a random number z1 in the interval (0, 1), determine the relative probability sorting interval position of the random number, and determine the component that first triggers the failure and the triggered failure mode.
[0055] In a specific embodiment, S400: analyzing the components that first trigger the failure and the failure mode triggered, and outputting the theoretical results of the fault type corresponding to the charging pile for new energy vehicles:
[0056] S410: Building a charging pile fault detection model;
[0057] S420: The component that first triggers the failure and the triggered failure mode are used as input to the charging pile fault detection model, and the theoretical results of the fault type corresponding to the charging pile for new energy vehicles are output.
[0058] Specifically, the charging pile fault detection model includes three layers: a data mapping layer, a feature extraction layer, and a classification recognition layer. The data mapping layer uses a convolutional neural network to map the one-dimensional time series signal of the operating status information in the sample library into a multidimensional time series signal according to the operating status in the sample library. Preferably, 1000 convolution kernels with a size of 1*100 can be established to map the one-dimensional time series signal of the operating status information into a 1000-dimensional time series signal, wherein the convolution step of the convolution kernel is 1; in order to keep the length of the data after convolution unchanged, the "SAME" convolution mode of adding 0 before and after the one-dimensional time series signal of the operating status information is adopted during convolution, and the number of zeros added before and after the one-dimensional time series signal is 100-1=99; the data size of the convolution kernel is 1*100, and its initial value is set to 0.01 (1 / 100), and the final value is obtained by the gradient descent (GDA) optimal estimation method.
[0059] The feature extraction layer includes establishing a deep neural network (DNN) and using mechanisms such as pooling and dropout to extract fault type features from multidimensional time series signals. Pooling is to reduce the data dimension, and the size of the pooling kernel is 3x3. In order to sharpen and highlight the data features, the maximum pooling method is adopted, that is, the maximum value in the 3x3 module is taken. Dropout is to prevent overfitting and increase the robustness of the system. Preferably, the probability of dropout is 0.8. The feature extraction layer has a total of 17 layers. This result is the most ideal data obtained by the staff after several experiments.
[0060] The classification and recognition layer includes establishing a fully connected layer and a dimension reduction output layer such as logistic regression combined with a label library label to realize fault type feature classification and recognition. The fully connected layer consists of three layers. The first layer is used to convert the multi-dimensional data of the feature output layer into one-dimensional data. The second and third layers are used to further extract the fault type features. The logistic regression layer converts the output of the fully connected layer into two values, 0 and 1. Preferably, it can be represented by y'. When the output y'=1, it indicates that the fault type feature extraction is accurate. When the output y'=0, it indicates that the fault type feature extraction is inaccurate.
[0061] In addition, in a specific embodiment, extracting the fault type features of the operating status from the multidimensional time series signal according to the pre-stored deep neural network includes:
[0062] According to the pre-stored deep neural network and using the pooling mechanism to reduce the dimension of the multi-dimensional time series signal. According to the pre-stored deep neural network and using the discarding mechanism to extract the fault type feature of the operating state of the multi-dimensional time series signal with reduced dimension.
[0063] Specifically, the dimension of the multidimensional time series signal is reduced by using a pooling mechanism according to a pre-stored deep neural network. The fault type feature is extracted from the multidimensional time series signal with reduced dimension according to a pre-stored deep neural network and a discarding mechanism. Pooling is to reduce the data dimension. The size of the pooling kernel is 3x3. In order to sharpen the data features, the maximum pooling method is adopted, that is, the maximum value in the 3x3 module is taken. Dropout is to prevent overfitting and increase the robustness of the system. Preferably, the probability of dropout is 0.8. There are 17 feature extraction layers in total. This result is the most ideal data obtained by the staff after several experiments.
[0064] In a specific embodiment, it also includes: S500: If the theoretical result of the fault type is inconsistent with the actual result of the fault type corresponding to the charging pile operating status in the pre-established sample library, the charging pile fault detection model is adjusted and updated according to the pre-stored gradient descent model, the actual result of the fault type and the neural network, and step S420 is re-executed.
[0065] On the other hand, see Appendix Figure 2 As shown, the present invention also discloses a charging pile charging failure analysis system based on data mining, comprising:
[0066] Acquisition module: used to obtain potential failure modes based on the operating characteristics of charging piles for new energy vehicles and combined with the failure status and records of current and previous processes;
[0067] A calculation module, connected to the acquisition module, is used to perform data mining of failure modes based on a DBSCAN clustering algorithm to obtain the number of failed components in the charging module of a charging pile for new energy vehicles;
[0068] The processing module is connected to the calculation module and is used to sort the relative probabilities corresponding to different failure modes of all components in the charging module of the charging pile for new energy vehicles according to the number of failed components, and determine the components that trigger the failure first and the triggered failure mode;
[0069] The analysis module is connected to the processing module and is used to analyze the components that first trigger failure and the triggered failure mode, and output the theoretical results of the fault type corresponding to the charging pile for new energy vehicles.
[0070] It can be seen from the above technical solutions that, compared with the prior art, the present invention discloses a charging pile charging failure analysis method and system based on data mining, which can efficiently and conveniently realize the failure analysis monitoring of the charging pile, and the specific beneficial effects are:
[0071] (1) The present invention first determines the component that first triggers the failure and the failure mode triggered, finds the root cause of the failure, and then obtains the corresponding fault type theoretical result. The process is simple and can quickly and accurately obtain the charging fault of a complex electronic control system such as a charging pile, thereby improving the efficiency of repairing a faulty charging pile.
[0072] (2) A neural network is used to establish charging pile fault type detection, and the theoretical results of the fault type are compared and optimized with the actual results of the fault type corresponding to the charging pile operating status in the sample library to make the detection results more accurate.
[0073] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.
[0074] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.
Claims
1. A charging pile charging failure analysis method based on data mining, based on a charging pile for new energy vehicles, characterized in that: The following steps are involved: S100: Based on the operating characteristics of the charging pile for new energy vehicles, combined with the failure status and records of the current and previous processes, obtain its potential failure mode; S200: Perform data mining of failure modes based on the DBSCAN clustering algorithm to obtain the number of failed components in the charging module of the charging pile for new energy vehicles; S300: sorting the relative probabilities corresponding to different failure modes of all components in the charging module of the charging pile for the new energy vehicle according to the number of failed components, and determining the component that first triggers the failure and the triggered failure mode; S400: Analyze the components that are triggered to fail first and the triggered failure modes, and output theoretical results of the fault types corresponding to the charging piles for new energy vehicles; Wherein, the S200 includes: S210: Count the parameters of three failure modes and standardize the parameters, including: insulation aging failure, charging interface contact wear failure, DC contactor adhesion failure, charging gun electronic lock failure, network communication failure, and cable assembly disconnection failure, to obtain a failed component data set; S220: setting neighborhood parameters (ε, MinPts), where ε describes the neighborhood distance threshold of a component sample, and MinPts describes the threshold of the number of samples in the neighborhood of a component sample whose distance is ε; using the DBSCAN clustering algorithm, classifying the component samples of the core points on the failed component data set, and analyzing the clustering results; S230: fixing the core points of the data of all the failed components with labels, classifying the samples of the core points on the failed component data set by using the DBSCAN clustering algorithm with adaptive feature weights, and obtaining clustering results; S240: Count the clustering results to obtain the number of failed components. The calculation formula is as follows: N(xi)={xj∈D|distance(xi,xj)≤Eps}; Wherein, N(xi) represents the set of failed component sample points in the Eps neighborhood of xi, and |N(xi)| is the number of samples in the failed component sample set; The S300 includes: S310: sorting the relative probabilities corresponding to different failure modes of all components in the charging module of the charging pile for new energy vehicles; S320: Generate a random number z1 in the interval (0, 1), determine the relative probability sorting interval position of the random number, and determine the component that first triggers the failure and the triggered failure mode.
2. A charging pile charging failure analysis method based on data mining according to claim 1, characterized in that: The potential failure modes include: performance degradation failure, functional failure and sudden failure of charging piles for new energy vehicles.
3. The charging pile charging failure analysis method based on data mining according to claim 1 is characterized in that: S400: Analyze the components that trigger the failure first and the failure mode that triggers it, and output the theoretical results of the fault type corresponding to the charging pile for new energy vehicles: S410: Building a charging pile fault detection model; S420: Taking the component that first triggers failure and the triggered failure mode as input to the charging pile fault detection model, and outputting the theoretical result of the fault type corresponding to the charging pile for new energy vehicles.
4. The charging pile charging failure analysis method based on data mining according to claim 3 is characterized in that: It also includes: S500: If the theoretical result of the fault type is inconsistent with the actual result of the fault type corresponding to the charging pile operating status in the pre-established sample library, the charging pile fault detection model is adjusted and updated according to the pre-stored gradient descent model, the actual result of the fault type and the neural network, and the S420 step is re-executed.
5. A charging pile charging failure analysis system based on data mining using the charging pile charging failure analysis method based on data mining according to any one of claims 1 to 4, characterized in that: include: Acquisition module: used to obtain potential failure modes based on the operating characteristics of charging piles for new energy vehicles and combined with the failure status and records of current and previous processes; A calculation module, connected to the acquisition module, is used to perform data mining of failure modes based on a DBSCAN clustering algorithm to obtain the number of failed components in the charging module of the charging pile for new energy vehicles; A processing module connected to the calculation module is used to sort the relative probabilities corresponding to different failure modes of all components in the charging module of the charging pile for new energy vehicles according to the number of failed components, and determine the components that trigger the failure first and the triggered failure mode; The analysis module is connected to the processing module and is used to analyze the components that are first triggered to fail and the triggered failure mode, and output theoretical results of the fault types corresponding to the charging piles for new energy vehicles.
Citation Information
Patent Citations
Multi-failure-model-integrated direct-current charging pile operation reliability evaluation method for electric vehicle
CN110782126A
Data processing method and device, computer readable storage medium and processor
CN112612820A
Charging pile fault element prediction method and system
CN112699605A
Data mining processing method suitable for electric vehicle charging fault
CN114490836A
Method and apparatus for evaluating the relative aging state of a device battery based on a device group-based serial variation of the device batteries
DE102021203865A1
