A safety warning method and device during electric vehicle charging

Abstract

The present invention provides a safety warning method and device during the charging process of an electric vehicle, and relates to the field of data processing technology. In the present invention, based on the analysis of the fault characteristics existing in the historical charging process of the electric vehicle, the actual voltage value, current difference, voltage change rate, current change rate, temperature difference and the number of times the emergency stop button is touched are determined to perform information fusion, construct a health level status prediction value and divide the warning level, and then use the generative adversarial network to perform deep learning based on the result to obtain an operation status prediction model, input the real-time collected charging data into the operation status prediction model for identification, and obtain the charging operation status warning level. The present invention not only has high warning accuracy, high recognition degree and fast recognition rate for the charging fault warning of electric vehicles, can timely identify fault problems, but also greatly promotes the development of electric vehicles and related industries.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a safety early warning method and device during the charging process of an electric vehicle. Background Art

[0002] Globally, environmental and energy crises are becoming increasingly prominent. Compared to traditional fuel vehicles, electric vehicles offer significant advantages in conserving oil resources and reducing carbon emissions, attracting significant attention from governments and automakers worldwide. However, spontaneous combustion during charging can be a serious obstacle to the development of the electric vehicle industry. Research has found that battery overheating is a major cause of spontaneous combustion during electric vehicle charging. Therefore, developing a temperature warning model for the electric vehicle charging process, providing real-time battery temperature monitoring and safety warnings, can ensure charging safety and contribute to the sustainable development of the electric vehicle industry.

[0003] However, the existing fault diagnosis of various electric vehicle charging processes mostly relies on expert experience, which is unable to detect problems in time and prevent losses in time. In addition, the fault diagnosis system of existing electric vehicle charging warning devices is still imperfect, and the accuracy of electric vehicle safety warnings is also poor. Summary of the Invention

[0004] The purpose of the present invention is to provide a safety warning method and device during the charging process of an electric vehicle to improve the above-mentioned problem. In order to achieve the above-mentioned purpose, the technical solution adopted by the present invention is as follows:

[0005] In a first aspect, the present application provides a safety warning method during the charging process of an electric vehicle, comprising:

[0006] Based on the license plate code of the electric vehicle, historical data and real-time data of the electric vehicle charging are respectively obtained, and first data and second data are obtained after preprocessing respectively, where the first data and the second data are both physical quantities that can represent a fault during the charging process;

[0007] Extracting evaluation indicators based on the first data, and determining corresponding weight values and membership degrees respectively;

[0008] Constructing a health level status prediction value based on the first data, the weight value, and the membership degree using a weighted average principle;

[0009] Based on the first data and the health level status prediction value, a generative adversarial network model is used for training to obtain an operation status prediction model;

[0010] The second data is input into the operation state prediction model to obtain a charging operation state warning level.

[0011] In a second aspect, the present application further provides a safety warning device for an electric vehicle during charging, comprising an acquisition module, an extraction module, a construction module, a training module, and an early warning module, wherein: the acquisition module is configured to acquire historical data and real-time data of electric vehicle charging based on the license plate code of the electric vehicle, and to obtain first data and second data respectively after preprocessing, wherein the first data and the second data are both physical quantities that can characterize a fault during the charging process;

[0012] Extraction module: used for extracting evaluation indicators based on the first data, and determining corresponding weight values and membership degrees respectively;

[0013] A construction module is configured to construct a health level status prediction value based on the first data, the weight value, and the membership degree using a weighted average principle;

[0014] A training module is configured to perform training using a generative adversarial network model based on the first data and the health level status prediction value to obtain an operation status prediction model;

[0015] Early warning module: used for inputting the second data into the operation status prediction model to obtain the charging operation status early warning level.

[0016] The beneficial effects of the present invention are:

[0017] The present invention analyzes the characteristics of historical charging failures in electric vehicles and identifies physical quantities in historical data that can characterize charging failures, such as actual voltage value, current difference, voltage change rate, current change rate, temperature difference, and the number of times the emergency stop button is pressed. After preprocessing (i.e., first data), these multi-dimensional evaluation indicators are fused to construct health level status prediction values and divide warning levels. Based on this result, a generative adversarial network is used for deep learning to obtain an operating status prediction model. Secondary data, obtained by preprocessing real-time charging data, is input into the operating status prediction model for identification, resulting in a charging operating status warning level. Compared to issuing warnings based on deviations from a preset value for a single evaluation indicator collected in real time, the present invention integrates information from various evaluation indicator data collected by monitoring electric vehicle charging to achieve charging fault warnings for electric vehicles. This not only effectively avoids false alarms caused by erroneous charging data, but also provides high accuracy, high recognition, and high speed warning level results, enabling timely identification of fault problems, significantly promoting the development of electric vehicles and related industries.

[0018] In the present invention, in the preprocessing method for charging data, outliers are removed based on the ID of the same type of data corresponding to the charging data, and then filled. While ensuring the integrity of the data, it also avoids the interference of data defects on the subsequent prediction results, thereby improving the accuracy of the early warning results. The complete data is then discretized using the time factor. The obtained discrete data is more accurate than the corresponding original data and more in line with the data feature input criteria of the operating status prediction model. The discretized data is substituted into the later operating status prediction model for operation, which can improve the operating performance of the model. Normalization is then performed to accelerate gradient descent, which can accelerate the convergence of the generative adversarial network model, shorten the time to find the optimal solution of the model, and also improve the accuracy of the model to prevent gradient explosion.

[0019] In constructing the health level status prediction value, considering that the constant weights of indicators assigned by the hierarchical analysis method will not change when a certain evaluation indicator deteriorates during the charging process, which will affect the accuracy of the evaluation of a large amount of charging data, the constant weights of the sub-item layer and the evaluation indicator layer are corrected respectively to solve the problem of variable degradation degree generated by charging data, so as to improve the accuracy of the early warning results.

[0020] Other features and advantages of the present invention will be described in the following description, and in part will become apparent from the description, or understood by practicing the embodiments of the present invention. The purposes and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description, claims, and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without paying any creative work.

[0022] Figure 1 This is a flow chart of a safety warning method during charging of an electric vehicle according to an embodiment of the present invention;

[0023] Figure 2 This is a schematic structural diagram of a safety warning device during charging of an electric vehicle according to an embodiment of the present invention;

[0024] Figure 3 Schematic diagram of the device structure of the safety warning method during the charging process of an electric vehicle described in an embodiment of the present invention.

[0025] In the figure: 710-acquisition module; 711-marking unit; 712-modification unit; 713-filling unit; 714-discrete unit; 7141-acquisition unit; 7142-transformation unit; 7143-multi-dimensional unit; 715-preprocessing unit; 720-extraction module; 721-evaluation unit; 722-judgment unit; 723-calculation unit; 724-judgment unit; 725-first statistical unit; 726-second statistical unit; 730-construction module; 740-training module; 741-learning unit; 750-warning module; 800-safety warning method and equipment; 801-processor; 802-memory; 803-multimedia component; 804-I / O interface; 805-communication component. DETAILED DESCRIPTION

[0026] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, 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 part of the embodiments of the present invention, not all of the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings herein can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0027] It should be noted that similar reference numerals and letters represent similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined or explained in subsequent drawings. At the same time, in the description of the present invention, the terms "first", "second", etc. are used only to distinguish the description and should not be understood as indicating or implying relative importance.

[0028] Example 1:

[0029] This embodiment provides a safety early warning method during the charging process of an electric vehicle.

[0030] See also Figure 1 , Figure 1 , the method includes steps S1, S2, S3, S4 and S5, wherein:

[0031] Step S1: Based on the license plate code of the electric vehicle, historical data and real-time data of the electric vehicle charging are obtained respectively, and first data and second data are obtained respectively after preprocessing. The first data and the second data are both physical quantities that can characterize a fault during the charging process.

[0032] It is understandable that in this step, the historical charging data of the electric vehicle in the database is used based on the unique code of the electric vehicle license plate, and the real-time charging data of the electric vehicle is collected through various sensors and information collection equipment in the charging pile. Then, the missing data, interference information, etc. in the historical data and real-time data are preprocessed respectively to ensure the integrity of the data while avoiding the interference of data defects on the later prediction results, thereby improving the accuracy of the early warning results. It should be noted that in the historical data and real-time data of this embodiment, when a fault occurs during the charging process, the actual voltage value, current difference, voltage change rate, current change rate, temperature difference and the number of times the emergency stop button is touched will change significantly, so the above data types are collected from the historical data and real-time data for preprocessing.

[0033] The above preprocessing method includes step S11, step S12, step S13, step S14 and step S15.

[0034] Step S11: Mark the historical data or the real-time data according to the data type to obtain a type data ID.

[0035] It is understood that in this step, different ID labels are used to mark different data types based on the historical data or the real-time data to obtain the type data ID. For example, the type data IDs corresponding to the actual voltage value, current difference, voltage change rate, current change rate, temperature difference, and number of times the emergency stop button is pressed are A, B, C, D, E, and F, respectively.

[0036] Step S12: Modify based on the time series and the type data ID to obtain feature sequences at different times in the same dimension.

[0037] It can be understood that in this step, the ID of each collected information is modified based on the time series for each type of data, resulting in a feature sequence for the same dimension at different times. For example, the type data IDs of the actual voltage values collected at the Nth second, the N+1th second, the N+2th second, the N+3th second, the N+4th second, and the N+5th second are A+1, A+2, A+3, A+4, and A+5, respectively. The feature sequence is composed of A+1, A+2, A+3, A+4, and A+5.

[0038] Step S13: Based on all the feature sequences, outliers are removed and missing values are filled to obtain complete data.

[0039] It is understood that in this step, outliers are removed according to the preset range for each type of data. Multi-order Lagrangian interpolation is then used to fill in the null values that occurred during data collection and the missing data after outlier removal. This completes the data set to ensure the integrity of the electric vehicle charging dataset, providing training data for the subsequent generative adversarial network model and improving the overall performance of the operating status prediction model.

[0040] Step S14: performing discrete processing based on the complete data to obtain discrete data.

[0041] It is understood that in this step, the continuous complete data is uniformly converted into segments of discretized intervals, resulting in more accurate discrete data than the corresponding original data. This makes it more consistent with the data feature input criteria of the operation status prediction model. Substituting the discretized data into the subsequent operation status prediction model can improve the model's performance. Discretization processing methods include equidistance discretization, equal frequency discretization, and K-means model discretization.

[0042] Furthermore, the above discrete method includes step S141, step S142 and step S143.

[0043] Step S141 : collecting data at standard time intervals based on the complete data to form a time series vector.

[0044] It can be understood that in this step, the parameters corresponding to the interval points are collected in the complete data based on the same time interval, and the time series vector is constructed based on the time sequence corresponding to the complete data at each interval point. For example, the time series vector corresponding to the actual voltage value is x={x m1 ,x m2 …,x mn}, where x is a time series vector; x mn It is the charging operation status data of the mth type at the nth time interval.

[0045] Step S142: obtaining different standard time series vectors based on the time series vector being changed by a time factor.

[0046] It can be understood that in this step, the time scale factor is introduced to discretize the time series vector according to formula (1), so that the standard time series vector under different time lengths can be freely selected, thereby improving the accuracy of discrete data acquisition.

[0047]

[0048] in: is the standard time series vector; τ is the time factor; m is the total number of time interval points; n is the number of time separation points in the time series; k is the charging operation status data of different data types; x mk It is the discrete charging operation state data corresponding to the m time intervals corresponding to the k-th charging operation state data.

[0049] Step S143: construct a composite vector based on the different standard time series vectors to obtain discrete data.

[0050] It can be understood that in this step, discrete data is constructed according to formula (2):

[0051]

[0052] Among them: composite vector is a 1×H order vector; τ is the time factor; M is the data type, i.e., the different dimensions of the data; d j is the standard time series vector; h m is the dimension vector; η m is the delay time vector. In order to ensure that the nonlinear relationship between the operating state data does not change, the dimension vector h is embedded in formula (2) m In order to compress the running status data without losing data, a delay time vector η is introduced m , obtaining discrete data based on multi-dimensional information fusion and delayed transmission can not only ensure the integrity of data transmission, but also improve the accuracy of early warning results.

[0053] Step S15: performing normalization processing based on the discrete data to obtain first data or second data accordingly.

[0054] It is understandable that in this step, the range and variation characteristics of different types of charging data may vary greatly. If the original data is used directly to train the generative adversarial network, it may introduce large data noise, resulting in a large deviation in the final warning result. After normalization, discrete data can be calculated in the same dimension, so that different types of charging data can have the same importance in the operation process of the generative adversarial network. After normalization, the charging data features can accelerate gradient descent, accelerate the convergence of the generative adversarial network model, shorten the time to find the optimal solution of the model, and also improve the accuracy of the model to prevent gradient explosion. Normalization methods include 0-1 normalization and Z-score normalization.

[0055] Step S2: extract evaluation indicators based on the first data, and determine corresponding weight values and membership degrees respectively.

[0056] It is understood that in this step, algorithms such as expert scoring, PCA principal component analysis, and Pearson correlation coefficient are used to determine the weight of each evaluation indicator based on its importance in the charging health assessment. Fuzzy membership analysis is also used to determine the probability of membership of each evaluation indicator to the corresponding set.

[0057] In detail, step S2 includes step S21 , step S22 , step S23 , step S24 , step S25 and step S26 .

[0058] Step S21: construct an operation status evaluation system based on the first data using the analytic hierarchy process.

[0059] It can be understood that in this step, a three-level structure model of the target layer, the criterion layer and the solution layer is constructed in order from top to bottom based on the preprocessed historical data using the hierarchical analysis method, wherein the target layer is the operating status evaluation system (corresponding to healthy, sub-healthy, abnormal and serious operating states, respectively), the criterion layer is the electrical evaluation index and the temperature evaluation index, etc., and the solution layer is the actual voltage value, current change rate and voltage change rate corresponding to the electrical evaluation index, etc., and the temperature difference corresponding to the temperature evaluation index, etc.

[0060] Step S22: constructing an indicator evaluation matrix based on the operating status evaluation system.

[0061] It can be understood that in this step, the evaluation indicators are compared layer by layer based on the three-level structure model to obtain the relationship of relative importance, and the nine-point scaling method is used to score each evaluation indicator to obtain the discriminant matrix. The discriminant matrix is shown in formula (3):

[0062] A=(a ij ) n×n (3)

[0063] Among them: A is the discriminant matrix; a ij It is divided into the importance of evaluation index i and evaluation index j of the current level of the discriminant matrix to the previous level; i and j are different types of evaluation indicators respectively; n is the dimension of the hierarchical model.

[0064] Step S23: Calculate the eigenvector and the maximum eigenvalue based on the index evaluation matrix.

[0065] Step S24: Calculate a consistency ratio based on the maximum eigenvalue; if the consistency ratio is less than 0.1, determine a weight value according to the eigenvector; otherwise, reallocate the weight value.

[0066] It can be understood that in this step, the consistency ratio of the measurement judgment matrix deviation from consistency is obtained based on the maximum eigenvalue and formula (4):

[0067]

[0068] Where: R is the consistency ratio; λ max is the maximum eigenvalue of the discriminant matrix; n is the order of the discriminant matrix; and E is the average random consistency index. When R < 0.1, the judgment matrix satisfies the consistency index, indicating that the weight distribution is reasonable. However, the constant weight value corresponding to the eigenvector of the charging data will not be adjusted due to changes in the characteristic indicators of the charging data, which will affect the accuracy of the evaluation of a large amount of charging data. According to formula (5), the constant weight value is corrected to solve the problem of variable degradation degree generated by the charging data, so as to improve the accuracy of the warning results:

[0069]

[0070] Among them: a ij and It is divided into the constant weight value and variable weight value of the evaluation index i and evaluation index j of the current level to the previous level in the discriminant matrix; i and j are different types of evaluation indicators respectively; n is the dimension of the hierarchical model; m is the number of data types under the i-th evaluation indicator; α is the weight adjustment rate index; λ ik If R≥0.1 does not meet the consistency ratio, the weight value needs to be redistributed.

[0071] Step S25: Determine a sample domain based on the first data, and determine an operating status set based on the sample domain; the operating status set is a range of operating data corresponding to different health levels.

[0072] It can be understood that in this step, the sample domain corresponding to the first data of the historical charging data after preprocessing is set as U, and any charging data sample U0∈U is selected. Then, according to a movable boundary of the charging data sample domain U, a variable charging pile set A is obtained. * .

[0073] Step S26: Perform statistics based on the sample domain and the operating state set respectively to obtain a membership degree, where the membership degree is a limit value of the number of times the operating state set occurs in the sample domain.

[0074] It can be understood that in this step, the membership degree is calculated according to formula (6):

[0075]

[0076] Where: μ is the membership degree; U0 is any charging data sample; A * A collection of variable charging piles.

[0077] Step S3: constructing a health level status prediction value based on the first data, the weight value, the membership degree and the weighted average principle.

[0078] It can be understood that in this step, the health index is calculated based on formula (7), and the health level is divided based on the health index as shown in Table 1:

[0079]

[0080] Where: HI is the health index matrix; D is the charging data dimension; ω i is the first data of the i-th dimension; a ij is the weight value corresponding to each evaluation index; μ ij The corresponding degree of membership for each evaluation indicator weight is denoted as follows: Assume that the health level is divided into four operating status levels: healthy, sub-healthy, abnormal, and severe:

[0081] Table 1 Health level classification table

[0082]

[0083] Step S4: Based on the first data and the health level status prediction value, a generative adversarial network model is used for training to obtain an operating status prediction model.

[0084] It can be understood that in this step, deep learning is used to learn the various evaluation indicators and health level status during the charging process, and the generative adversarial network model is trained based on the first data corresponding to the time series to obtain an operation status prediction model, so as to realize real-time prediction of the charging data corresponding to the next moment, so as to improve the recognition rate of the health level.

[0085] In detail, the training method of the operation status prediction model includes step S41: based on the first data and the health level status prediction value, the generator model and the discriminator model of the generative adversarial network model are trained alternately, wherein a training cycle process of the alternating training is: taking the first data as the input value of the generator model to obtain generated data; then taking the generated data and the health level status prediction value as the input value of the discriminator model, using the loss value and the residual value fed back by the discriminator model to update the weights, minimizing the residual value and the loss value, and obtaining the operation status prediction model, wherein the residual value is the Wasserstein distance between the health level status prediction value and the generated data. In the present invention, the Wasserstein distance is used as the distance metric for equivalent optimization to solve the problems of gradient vanishing, gradient instability and training model instability in the generative adversarial network model due to the unreasonable equivalent optimization distance metric method.

[0086] Step S5: input the second data into the operation status prediction model to obtain a charging operation status warning level.

[0087] It can be understood that in this step, inputting the second data into the operating status prediction model can predict the health level operating status corresponding to the next moment in real time, and obtain the corresponding charging operating status warning level according to the warning level classification in Table 1.

[0088] Example 2:

[0089] like Figure 2 The present embodiment provides a safety warning device for an electric vehicle during charging, including an acquisition module 710, an extraction module 720, a construction module 730, a training module 740, and a warning module 750, wherein:

[0090] Acquisition module 710: used to obtain historical data and real-time data of electric vehicle charging based on the license plate code of the electric vehicle, and obtain first data and second data respectively after preprocessing. The first data and the second data are both physical quantities that can characterize faults occurring during the charging process.

[0091] Optionally, the acquisition module 710 includes a marking unit 711, a modification unit 712, a filling unit 713, a discretization unit 714, and a pre-processing unit 715, wherein:

[0092] The marking unit 711 is configured to mark the historical data or the real-time data according to the data type to obtain a data type ID;

[0093] Modification unit 712: used for performing modification based on the time series and the type data ID to obtain feature sequences at different times in the same dimension;

[0094] Filling unit 713: used for removing outliers based on all the feature sequences and then filling missing values to obtain complete data;

[0095] Discrete unit 714: configured to perform discrete processing based on the complete data to obtain discrete data.

[0096] Furthermore, the discrete unit 714 includes a collection unit 7141, a transformation unit 7142, and a multi-dimensional unit 7143, wherein:

[0097] The acquisition unit 7141 is configured to acquire the complete data using a standard time interval to form a time series vector;

[0098] Transformation unit 7142: configured to transform the time series vector by a time factor to obtain different standard time series vectors;

[0099] Multi-dimensional unit 7143: used to construct a composite vector based on the different standard time series vectors to obtain discrete data.

[0100] The pre-processing unit 715 is configured to perform normalization processing based on the discrete data to obtain the first data or the second data accordingly.

[0101] Extraction module 720: used to extract evaluation indicators based on the first data and determine corresponding weight values and membership degrees respectively.

[0102] Optionally, the extraction module 720 includes an evaluation unit 721, a judgment unit 722, a calculation unit 723, a judgment unit 724, a first statistics unit 725, and a second statistics unit 726, wherein:

[0103] Evaluation unit 721: used for constructing an operation status evaluation system based on the first data by using the analytic hierarchy process;

[0104] Evaluation unit 722: configured to construct an indicator evaluation matrix based on the operation status evaluation system;

[0105] Calculation unit 723: used to calculate the eigenvector and the maximum eigenvalue based on the index evaluation matrix;

[0106] A judgment unit 724 is configured to calculate a consistency ratio based on the maximum eigenvalue, and if the consistency ratio is less than 0.1, determine a weight value according to the eigenvector; otherwise, reallocate the weight value;

[0107] A first statistical unit 725 is configured to determine a sample domain based on the first data, and determine an operating state set based on the sample domain; the operating state set is a range of operating data corresponding to different health levels;

[0108] The second statistical unit 726 is configured to perform statistics based on the sample domain and the operating state set respectively to obtain a membership degree, where the membership degree is a limit value of the number of times the operating state set occurs in the sample domain.

[0109] Construction module 730: configured to construct a health level status prediction value based on the first data, the weight value, and the membership degree using a weighted average principle;

[0110] Training module 740: used to train the generative adversarial network model based on the first data and the health level status prediction value to obtain an operation status prediction model.

[0111] In detail, the training module 740 includes a learning unit 741, wherein:

[0112] Learning unit 741: used to perform alternating training on the generator model and the discriminator model of the generative adversarial network model based on the first data and the health level status prediction value, respectively, wherein a training cycle process of the alternating training is: taking the first data as the input value of the generator model to obtain generated data; then taking the generated data and the health level status prediction value as the input values of the discriminator model, using the loss value and the residual value fed back by the discriminator model to update the weights, minimizing the residual value and the loss value, and obtaining an operation status prediction model, wherein the residual value is the Wasserstein distance between the health level status prediction value and the generated data.

[0113] The early warning module 750 is configured to input the second data into the operation status prediction model to obtain an early warning level of the charging operation status.

[0114] It should be noted that, regarding the apparatus in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated on here.

[0115] Example 3:

[0116] Corresponding to the above method embodiment, this embodiment also provides a safety warning method device 800 during the charging process of an electric vehicle. The safety warning method device 800 during the charging process of an electric vehicle described below and the safety warning method during the charging process of an electric vehicle described above can be referenced to each other.

[0117] Figure 3 FIG. 8 is a block diagram of a safety warning method and device 800 for an electric vehicle charging process according to an exemplary embodiment. Figure 3 As shown, the electric vehicle charging safety warning method device 800 may include: a processor 801, a memory 802. The electric vehicle charging safety warning method device 800 may also include one or more of a multimedia component 803, an I / O interface 804, and a communication component 805.

[0118] The processor 801 is used to control the overall operation of the electric vehicle charging safety warning method device 800 to complete all or part of the steps of the above-mentioned electric vehicle charging safety warning method. The memory 802 is used to store various types of data to support the operation of the electric vehicle charging safety warning method device 800. This data may include, for example, instructions for any application or method operating on the electric vehicle charging safety warning method device 800, as well as application-related data such as contact data, sent and received messages, pictures, audio, video, etc. The memory 802 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk. The multimedia component 803 may include a screen and an audio component. The screen may be, for example, a touch screen, and the audio component is used to output and / or input audio signals. For example, the audio component may include a microphone for receiving external audio signals. The received audio signal may be further stored in the memory 802 or transmitted via the communication component 805. The audio component also includes at least one speaker for outputting audio signals. The I / O interface 804 provides an interface between the processor 801 and other interface modules, which may be a keyboard, a mouse, buttons, etc. These buttons may be virtual buttons or physical buttons. The communication component 805 is used for wired or wireless communication between the safety warning method device 800 and other devices during the electric vehicle charging process. Wireless communication, such as Wi-Fi, Bluetooth, Near Field Communication (NFC), 2G, 3G or 4G, or a combination of one or more thereof, therefore, the corresponding communication component 805 may include: a Wi-Fi module, a Bluetooth module, an NFC module.

[0119] In an exemplary embodiment, the safety warning method device 800 during the charging process of an electric vehicle can be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to execute the above-mentioned safety warning method during the charging process of an electric vehicle.

[0120] In another exemplary embodiment, a computer storage medium including program instructions is further provided. When executed by a processor, the program instructions implement the steps of the above-described safety warning method for an electric vehicle charging process. For example, the computer storage medium may be the aforementioned memory 802 including the program instructions. The program instructions may be executed by the processor 801 of the apparatus 800 for the safety warning method for an electric vehicle charging process to implement the above-described safety warning method for an electric vehicle charging process.

[0121] Example 4:

[0122] Corresponding to the above method embodiment, this embodiment further provides a storage medium. The storage medium described below and the safety warning method during charging of an electric vehicle described above can refer to each other.

[0123] A storage medium stores a computer program, which, when executed by a processor, implements the steps of the safety warning method during charging of an electric vehicle in the above-mentioned method embodiment.

[0124] The storage medium may specifically be any storage medium capable of storing program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0125] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Persons skilled in the art will readily appreciate that the present invention is susceptible to various modifications and variations. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention are intended to be within the scope of protection of the present invention.

[0126] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be readily conceived by a person skilled in the art within the technical scope disclosed in this invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A safety warning method for an electric vehicle during charging, characterized in that: include: Based on the license plate code of the electric vehicle, historical data and real-time data of the electric vehicle charging are respectively obtained, and first data and second data are obtained after preprocessing respectively, where the first data and the second data are both physical quantities that can represent a fault during the charging process; Extracting evaluation indicators based on the first data, and determining corresponding weight values and membership degrees respectively; Based on the first data, the weight value, and the degree of membership, a weighted average principle is used to construct a health level status prediction value; Based on the first data and the health level status prediction value, a generative adversarial network model is used for training to obtain an operation status prediction model; Inputting the second data into the operation state prediction model to obtain a charging operation state warning level; Extracting evaluation indicators based on the first data and determining corresponding weight values and membership degrees respectively include: Based on the first data, an operation status evaluation system is constructed using the analytic hierarchy process; Construct an indicator evaluation matrix based on the operating status evaluation system; Calculate the eigenvector and maximum eigenvalue based on the index evaluation matrix; Calculate the consistency ratio based on the maximum eigenvalue, and if the consistency ratio is less than 0.1, determine the weight value according to the eigenvector; otherwise, reallocate the weight value; Determining a sample domain based on the first data, and determining an operating state set based on the sample domain; the operating state set is a range of operating data corresponding to different health levels; Based on the statistics of the sample domain and the running state set, the membership degree is obtained. The membership degree is the limit value of the number of times the running state set occurs in the sample domain. Based on the first data and the health level status prediction value, the generative adversarial network model is used for training to obtain the operating status prediction model including: Based on the first data and the health level status prediction value, the generator model and the discriminator model of the generative adversarial network model are trained alternately, respectively. A training cycle process of the alternating training is: the first data is used as the input value of the generator model to obtain the generated data; then the generated data and the health level status prediction value are used as the input values of the discriminator model, and the loss value and residual value fed back by the discriminator model are used to update the weights, and the residual value and the loss value are minimized to obtain the operation status prediction model, and the residual value is the Wasserstein distance between the health level status prediction value and the generated data.

2. The safety warning method during charging of an electric vehicle according to claim 1, characterized in that: The pretreatment method comprises: Based on historical data or real-time data, mark them according to the data type to obtain the type data ID; Modify based on time series and type data ID to obtain feature sequences at different times in the same dimension; Based on all feature sequences, outliers are removed and missing values are filled to obtain complete data; Discrete data is obtained by performing discrete processing based on complete data; Normalization processing is performed based on the discrete data to obtain the first data or the second data accordingly.

3. The safety warning method during charging of an electric vehicle according to claim 2, characterized in that: Discrete processing is performed based on the complete data, and the discrete data obtained include: Based on the complete data, the data is collected at standard time intervals to form a time series vector; Based on the change of time series vector through time factor, different standard time series vectors are obtained; Composite vectors are constructed based on different standard time series vectors to obtain discrete data.

4. A safety warning device for electric vehicle charging, characterized in that: include: Acquisition module: used to acquire historical data and real-time data of electric vehicle charging based on the license plate code of the electric vehicle, and obtain first data and second data respectively after preprocessing, wherein the first data and the second data are both physical quantities that can represent a fault in the charging process; Extraction module: used for extracting evaluation indicators based on the first data and determining corresponding weight values and membership degrees respectively; Construction module: used for constructing a health level status prediction value based on the first data, the weight value, and the membership degree using a weighted average principle; A training module is configured to perform training using a generative adversarial network model based on the first data and the health level status prediction value to obtain an operation status prediction model; An early warning module is configured to input the second data into an operation state prediction model to obtain an early warning level of the charging operation state; The extraction module includes: Evaluation unit: used for constructing an operation status evaluation system based on the first data by using the analytic hierarchy process; Evaluation unit: used to construct an indicator evaluation matrix based on the operating status evaluation system; Calculation unit: used to calculate the eigenvector and the maximum eigenvalue based on the index evaluation matrix; Judgment unit: used to calculate the consistency ratio based on the maximum eigenvalue. If the consistency ratio is less than 0.1, the weight value is determined according to the eigenvector; otherwise, the weight value is reallocated. A first statistical unit is configured to determine a sample domain based on the first data, and determine an operating state set based on the sample domain; the operating state set is a range of operating data corresponding to different health levels; The second statistical unit is used to perform statistics based on the sample domain and the operating state set respectively to obtain the membership degree, which is the limit value of the number of times the operating state set occurs in the sample domain; Training modules include: Learning unit: used to perform alternating training on the generator model and the discriminator model of the generative adversarial network model based on the first data and the health level status prediction value, wherein a training cycle process of the alternating training is: taking the first data as the input value of the generator model to obtain the generated data; then taking the generated data and the health level status prediction value as the input value of the discriminator model, using the loss value and the residual value fed back by the discriminator model to update the weights, minimizing the residual value and the loss value, and obtaining the operation status prediction model, wherein the residual value is the Wasserstein distance between the health level status prediction value and the generated data.

5. The safety warning device for electric vehicle charging according to claim 4, characterized in that: The acquisition module includes: Marking unit: used to mark historical data or real-time data according to data type to obtain type data ID; Modification unit: used to modify based on time series and type data ID to obtain feature sequences at different times in the same dimension; Filling unit: used to remove outliers based on all feature sequences and then fill in missing values to obtain complete data; Discrete unit: used to perform discrete processing based on complete data to obtain discrete data; Preprocessing unit: used to perform normalization processing based on discrete data to obtain first data or second data accordingly.

6. The safety warning device for electric vehicle charging according to claim 5, characterized in that: Discrete elements include: Acquisition unit: used to collect complete data using standard time intervals to form a time series vector; Transformation unit: used to change the time series vector through the time factor to obtain different standard time series vectors; Multidimensional unit: used to construct composite vectors based on different standard time series vectors to obtain discrete data.

Images