Electromobile load monitoring method and system based on EMD feature extraction and SVM

Abstract

The invention discloses a battery car load monitoring method based on empirical mode decomposition (EMD) feature extraction and a support vector machine (SVM), and the method comprises the steps: extracting user current original data through a user intelligent terminal; identifying an equipment transient starting state based on the current data, and extracting an equipment starting event based on a sliding window; extracting voltage and current wavelet features based on EMD (Empirical Mode Decomposition) transformation; and judging the battery car event based on a support vector machine. According to the method, the problem that the indoor charging of the battery car is difficult to identify in reality is solved, and a solution is provided for reducing fire potential safety hazards caused by the indoor charging of the battery car.

Description

Technical Field

[0001] The present invention belongs to the field of power distribution and utilization, and relates to a method for improving load identification in a non-intrusive field, and specifically relates to a method and system for monitoring the load of an electric battery vehicle based on EMD feature extraction and SVM. Background Art

[0002] With the rapid progress of social industrialization, the traditional power industry is developing in a more intensive and intelligent direction, and the importance of distribution-side demand has become stronger. The advancement of urbanization has led to an increase in community population density year by year, and the management of the distribution side has become increasingly difficult, especially the management of electric bicycles. According to statistics, the number of electric bicycles in the country exceeds 250 million (annual growth rate of 10%). However, due to improper use, about 2,600 electric vehicle fires occur each year. Entering buildings and households, batteries entering households, people and vehicles in the same room, and charging with flying wires have caused serious social harm to the society. Therefore, the management of the power demand side is crucial. How to effectively identify the charging status of household electric bicycles, understand user charging behavior, and guide users to consciously charge safely at charging piles is the key task of load-side management.

[0003] However, in actual scenarios, electric vehicles are small electrical devices with a power usually between 100-300W, and it is difficult to identify them through simple user power usage. In addition, during the identification process, the characteristics of the equipment are similar to those of power electronic equipment. For example, current technical documents provide VI curve detection technology to extract features of power loads, but the shortcoming of the existing technology is that electric vehicles are similar to power electronic equipment, and their VI images are similar, making it difficult to ensure the accuracy of feature recognition.

[0004] Therefore, there is an urgent need for a battery vehicle load monitoring method to explore charging characteristics and improve device identification accuracy. Summary of the invention

[0005] In order to solve the deficiencies in the prior art, the present invention provides a battery vehicle load monitoring method and system based on EMD feature extraction and SVM to mine charging characteristics and improve equipment identification accuracy.

[0006] The present invention adopts the following technical solution.

[0007] In a first aspect of the present invention, a method for monitoring battery vehicle load based on EMD feature extraction and SVM is provided, comprising the following steps:

[0008] Step 1, collecting the user's historical voltage original signal data and current original signal data;

[0009] Step 2: determine whether there is a transient device turn-on based on the current data, and if so, extract the turn-on event data based on the sliding window;

[0010] Step 3, based on the start-up event data, voltage and current wavelet features are extracted through EMD transformation to construct a load characteristic data set;

[0011] Step 4: Use a support vector machine to identify the load feature data set to determine whether a battery vehicle power load is generated.

[0012] Preferably, in step 1, the user's historical voltage original signal and current original signal are collected through the user's intelligent terminal, wherein the terminal sampling frequency is 6.4 khz.

[0013] Preferably, step 2 includes: extracting load current waveform data from the original current signal data, judging whether a transient start-up event of the equipment occurs by the extreme value change of the load current waveform data, when the extreme value change is greater than a threshold, an event occurs, otherwise no event occurs; based on the mutation principle, a sliding window is used to extract the data in the window as start-up event data.

[0014] Preferably, step 2 specifically includes:

[0015] Step 2.1, extracting load current waveform data I based on historical current original signal data;

[0016] Step 2.2, determine whether the extreme value of the load current waveform data has a sudden change, if so, execute step 2.3, where the mutation principle is: take w as a sliding window, if the difference between the current waveform data in the two windows is greater than the set threshold, an electrical equipment start event occurs;

[0017] Step 2.3: extract the data in the sliding window as the power-consuming device startup event data.

[0018] Preferably, the two windows are I s with I s+1 , the mutation principle formula is as follows:

[0019] I s =[i1, i2, …, i w ]

[0020] I s，α = sort(I s )(0.75·n)

[0021] I s+1.α -I s，α >TH

[0022] In the formula, i w represents the wth data in a sliding window of size w, sort is a function for arranging from small to large, I s，α is the third quartile, and TH is the event mutation threshold.

[0023] Preferably, step 3 includes decomposing the voltage original signal and the current original signal corresponding to the electrical equipment start-up event data, obtaining multi-level decomposition to generate intrinsic mode functions, respectively calculating the intrinsic mode function energy and the intrinsic mode function steepness of the intrinsic mode functions of different levels and establishing a feature data set.

[0024] Preferably, step 3 specifically includes:

[0025] Step 3.1, respectively decompose the voltage original signal and current original signal corresponding to the power-on event data of the power-consuming equipment extracted in step 2, and obtain the intrinsic mode function S generated by multi-level decomposition I :

[0026]

[0027] Where n is the number of intrinsic mode functions obtained by decomposition, c j is the jth intrinsic mode function, r is the residual signal;

[0028] Step 3.2, respectively calculate the intrinsic mode function energy and intrinsic mode function steepness of the intrinsic mode functions of different levels and establish a characteristic data set.

[0029] Preferably, the intrinsic mode function energy calculation formula is:

[0030]

[0031] In the formula, N is the number of data points contained in the sample, c j is the jth eigenmode function, c j (i) is the i-th data point of the j-th eigenmode function;

[0032] Steepness of the intrinsic mode function:

[0033]

[0034] In the formula, mean is the average value, μ is j is the mean value of the jth effective intrinsic mode function, σ j is the standard deviation of the jth effective eigenmode function.

[0035] Preferably, step 4 includes: dividing the data feature set into a training set and a test set, taking the battery vehicle start-up event as the positive sample X1 and the other electrical equipment start-up events as the negative sample X2, constructing a hyperplane to identify the load data, and when the result is in the positive sample interval of the hyperplane, it is represented as the battery vehicle being turned on, and when it is in the negative sample interval, it is represented as other electrical equipment.

[0036] Preferably, in step 4.1, the data feature set in step 3 is divided into a training set Ω Train and the test set ΩTest , the training set Ω Train The battery vehicle start-up event is taken as the positive sample X1, and the other electrical equipment start-up events are taken as the negative sample X2;

[0037] Step 4.2, construct a hyperplane H:w·x+b=0 to perform regression analysis on the data of different power-consuming equipment startup events, so that the two types of data X1 and X2 in the data set are separated, where w is the normal vector of the hyperplane and b represents the classification threshold, specifically:

[0038]

[0039] In the formula, ε i is the penalty factor; ξ is the penalty parameter used to penalize misclassification; x i is the eigenvalue of the i-th sample; n1 is the number of samples X1, n2 is the number of samples X2; minf(w, ε) is the minimum value of the target optimization function;

[0040] Step 4.3, based on the test set Ω Test Optimize the normal vector and classification threshold of the hyperplane H to obtain the optimal plane model, input the data feature set obtained in real time into the optimal plane model, and when the result is in the positive sample interval of the hyperplane, it is represented as an electric vehicle, and it is judged that the electric vehicle power load is generated in the power load data of the user; when it is in the negative sample interval, it is represented as other electrical equipment.

[0041] In a second aspect of the present invention, a battery vehicle load monitoring system based on EMD feature extraction and SVM is provided, the system comprising:

[0042] A data acquisition module is used to collect the user's original voltage signal data and current original signal data;

[0043] An event extraction module, used to determine the transient start of the device based on the current data, and extract the start event based on the sliding window;

[0044] A feature extraction module, used to extract current wavelet features through a fast wavelet transform algorithm based on the turn-on event data to obtain a data feature set;

[0045] The result judgment module is used to judge the battery vehicle event through the support vector machine based on the data feature set.

[0046] The beneficial effect of the present invention lies in that, compared with the prior art, high-dimensional feature extraction is performed based on the original current data, voltage and current wavelet features are extracted based on EMD transformation, and electric battery vehicles are identified in this feature dimension, thereby greatly improving the accuracy of identification; current waveform characteristics are obtained based on real-time collected current data of user electricity consumption, and whether a user power-on event occurs is determined by identifying whether a sudden change occurs in the current, and the electric battery vehicle start-up event and other power-on events are further divided through the optimized optimal plane model, providing timely, effective and accurate basis for identifying electric battery vehicles entering households for charging. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 It is the overall flow chart of the present invention.

[0048] Figure 2 This is the transient waveform of the battery vehicle being turned on. DETAILED DESCRIPTION

[0049] In order to make the purpose, technical scheme and advantages of the present invention clearer, the technical scheme of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. The embodiments described in this application are only part of the embodiments of the present invention, not all of them. Based on the spirit of the present invention, other embodiments obtained by ordinary technicians in this field without creative work are all within the scope of protection of the present invention.

[0050] In a first aspect of the present invention, a method for monitoring battery vehicle load based on EMD feature extraction and SVM is provided, comprising the following steps:

[0051] Step 1, collecting the user's historical voltage original signal data and current original signal data; preferably, in step 1, the user's historical voltage and current original signals are collected through the user's intelligent terminal, wherein the terminal sampling frequency is 6.4khz.

[0052] Step 2, based on the current data, determine whether there is a transient start of the equipment, and if so, extract the start event data based on the sliding window; Step 2 includes: extracting load current waveform data from the original current signal data, and judging whether a transient start event of the equipment occurs by the extreme value change of the load current waveform data. When the extreme value change is greater than the threshold, an event occurs, otherwise no event occurs; based on the mutation principle, the sliding window extracts the data in the window as the start event data.

[0053] Preferably, step 2 specifically includes:

[0054] Step 2.1, extracting load current waveform data I based on historical current original signal data;

[0055] Step 2.2, determine whether the extreme value of the load current waveform data has a sudden change, if so, execute step 2.3, wherein the sudden change principle is: take w as a sliding window, if the difference of the current waveform data in the two windows is greater than the set threshold, an electrical equipment start event occurs; preferably, the two windows are I s with I s+1 , the mutation principle formula is as follows:

[0056] I s =[i1, i2, …, i w ]

[0057] I s，α = sort(I s )(0.75·n)

[0058] I s+1.α -I s，α >TH

[0059] In the formula, i w represents the wth data in a sliding window of size w, sort is a function for arranging from small to large, I s，α is the third quartile, and TH is the event mutation threshold.

[0060] Step 2.3: extract the data in the sliding window as the power-consuming device startup event data.

[0061] Step 3, extracting current wavelet features through EMD transformation based on the start-up event data and constructing a load feature data set;

[0062] Step 3 includes respectively decomposing the voltage original signal and the current original signal corresponding to the power-on event data of the electrical equipment, obtaining multi-level decomposition to generate intrinsic mode functions, respectively calculating the intrinsic mode function energy and the intrinsic mode function steepness of the intrinsic mode functions of different levels and establishing a feature data set.

[0063] Step 3 specifically includes:

[0064] Step 3.1, respectively decompose the voltage original signal and current original signal corresponding to the power-on event data of the power-consuming equipment extracted in step 2, and obtain the intrinsic mode function S generated by multi-level decomposition I :

[0065]

[0066] Where n is the number of intrinsic mode functions obtained by decomposition, c j is the jth intrinsic mode function, r is the residual signal;

[0067] Step 3.2, respectively calculate the intrinsic mode function energy and intrinsic mode function steepness of the intrinsic mode functions of different levels and establish a characteristic data set.

[0068] The formula for calculating the energy of the eigenmode function is:

[0069]

[0070] In the formula, N is the number of data points contained in the sample, c j is the jth eigenmode function, c j (i) is the i-th data point of the j-th eigenmode function;

[0071] Steepness of the intrinsic mode function:

[0072]

[0073] In the formula, mean is the average value, μ is j is the mean value of the jth effective intrinsic mode function, σ j is the standard deviation of the jth effective eigenmode function.

[0074] Step 4: Use a support vector machine to identify the load feature data set to determine whether a battery vehicle power load is generated.

[0075] Step 4 includes: dividing the data feature set into a training set and a test set, taking the battery vehicle start-up event as the positive sample X1 and the other electrical equipment start-up events as the negative sample X2, constructing a hyperplane to identify the load data, and when the result is in the positive sample interval of the hyperplane, it is represented as the battery vehicle being turned on, and when it is in the negative sample interval, it is represented as other electrical equipment.

[0076] Preferably, in step 4.1, the data feature set in step 3 is divided into a training set Ω Train and the test set Ω Test , the training set Ω Train The battery vehicle start-up event is taken as the positive sample X1, and the other electrical equipment start-up events are taken as the negative sample X2;

[0077] Step 4.2, construct a hyperplane H:w·x+b=0 to perform regression analysis on the data of different power-consuming equipment startup events, so that the two types of data X1 and X2 in the data set are separated, where w is the normal vector of the hyperplane and b represents the classification threshold, specifically:

[0078]

[0079] In the formula, ε i is the penalty factor; ξ is the penalty parameter used to penalize misclassification; x i is the eigenvalue of the i-th sample; n1 is the number of samples X1, n2 is the number of samples X2; min f(w,ε) is the minimum value of the target optimization function;

[0080] Step 4.3, based on the test set Ω Test Optimize the normal vector and classification threshold of the hyperplane H to obtain the optimal plane model, input the data feature set obtained in real time into the optimal plane model, and when the result is in the positive sample interval of the hyperplane, it is represented as an electric vehicle, and it is judged that the electric vehicle power load is generated in the power load data of the user; when it is in the negative sample interval, it is represented as other electrical equipment.

[0081] In a second aspect of the present invention, a battery vehicle load monitoring system based on EMD feature extraction and SVM is provided, the system comprising:

[0082] Data acquisition module, used to collect user voltage signal and current raw signal data;

[0083] An event extraction module, used to determine the transient start of the device based on the current data, and extract the start event based on the sliding window;

[0084] A feature extraction module is used to extract voltage and current wavelet features based on the start-up event data through a fast wavelet transform algorithm to obtain a data feature set;

[0085] The result judgment module is used to judge the battery vehicle event through the support vector machine based on the data feature set.

[0086] Example 1: Figure 1 As shown, a battery vehicle load monitoring method based on EMD feature extraction and SVM is provided, comprising the following steps:

[0087] Step 1, collecting the user's historical voltage signal and historical current original signal data;

[0088] Preferably, in step 1, the user voltage signal U and the current original signal I are collected through the user intelligent terminal or the current and voltage transformer, and the terminal sampling frequency is preferably 6.4khz; step 1 also includes obtaining a variety of battery vehicles and other equipment load current waveform data samples, wherein the battery vehicle data set sample is Ω1, and the other equipment data set sample is Ω2. It can be understood that the collected user voltage signal U and the current original signal I include a variety of battery vehicles and other equipment load current waveform data samples. The present application determines the transient start of the equipment based on the historical voltage and current signal data, and extracts the start event based on the sliding window, that is, extracts the battery vehicle start event and other household appliance start events.

[0089] Step 2: determine whether the device is transiently turned on based on the current data, and extract the turn-on event based on the sliding window;

[0090] Step 2 includes: judging the occurrence of an event by the extreme value change of the load current sinusoidal waveform data; when the extreme value change is greater than a threshold, an event occurs; otherwise, no event occurs; Step 2.1: extracting the load current waveform data I, by Figure 2 As shown, it is the current data of the battery vehicle start-up event. After the event occurs, the extreme value of the current sine waveform changes suddenly. The principle of judging the sudden change is shown in step 2.2; the threshold setting principle is based on the various battery vehicles and other equipment load current waveform data samples collected in step 1.

[0091] Step 2.2: Assuming the current data is I, use the sliding window size w as the two windows divided into I s with I s+1 , and its mutation principles are as follows:

[0092] I s =[i1,i2,…,i n ]

[0093] I sort = sort(I s )

[0094] I s,α =I sort (0.75·n)

[0095] I s+1,α -I s,α >TH

[0096] In the formula, sort arranges functions from small to large, I s,α is the third quartile from the smallest to the largest, TH is the event mutation threshold, which is set to 0.5A here considering the common household devices;

[0097] Step 2.3: When step 2.2 is met, extract the data in the window as the start event.

[0098] In step 3, EMD features are extracted based on event data and the corresponding feature data set is obtained.

[0099] Step 3.1: Extract the turn-on event data based on step 2, assuming that the event e voltage data is U e , current data is I e ;

[0100] Step 3.2: Decompose the original signal based on the EMD algorithm to obtain multi-level decomposition to generate the intrinsic mode function (IMF);

[0101]

[0102] Where n is the number of IMFs obtained by decomposition, c j is the jth IMF, r is the residual signal, representing the average trend of the signal.

[0103] Step 3.3: Based on different levels of IMF, establish a feature data set. The feature content is as follows:

[0104] IMF Energy:

[0105]

[0106] Where N is the number of data points contained in the sample, m is the j-th IMF, and cj(i) is the i-th data point of the j-th IMF.

[0107] IMF steepness:

[0108]

[0109] In the formula, mean is the average value, μ is j is the mean of the jth effective IMF, σ j is the standard deviation of the j-th effective IMF.

[0110] Step 3.4: The content of the feature set is shown in Table 1.

[0111] Table 1

[0112]

[0113] Step 4: Based on the data feature set, the support vector machine is used to judge the battery vehicle event.

[0114] Based on the data feature set, the battery vehicle event is judged by SVM, and the specific steps are as follows:

[0115] Step 4.1: Randomly select 50% of the data feature set as the training set Ω Train , 50% as the test set Ω Test , the training set Ω Train The start-up event of the electric vehicle is taken as the positive sample X1, and the start-up events of other devices are taken as the negative sample X2.

[0116] Step 4.2: Construct a hyperplane H:w·x+b=0 to separate the two types of data X1 and X2. w is the normal vector of the optimal plane, and b is the classification threshold. Specifically:

[0117]

[0118] In the formula, ε i is the penalty factor, because during training, it cannot be guaranteed that every positive sample is actually a positive sample; ξ is the penalty parameter, which is used to penalize wrong classification.

[0119] Step 4.3: Based on Ω Train The obtained hyperplane H pairs the samples to be identified as the test set Ω Test When the result is in the positive sample interval of the hyperplane, it is represented as an electric vehicle and marked as 1. When it is in the negative sample interval, it is represented as a non-electric vehicle and marked as 0.

[0120] The beneficial effect of the present invention is that, compared with the prior art, high-dimensional feature extraction is performed based on the original voltage and current data, voltage and current wavelet features are extracted based on EMD transformation, and the battery vehicle is identified in this feature dimension, thereby greatly improving the accuracy of identification; the current waveform features are obtained based on the current data of user electricity consumption collected in real time, and whether the user's power-on event occurs by identifying whether the current has a sudden change, and further dividing the battery vehicle start-up event and other power-on events through the optimized optimal plane model, providing a timely, effective and accurate basis for identifying the battery vehicle charging at home. .

[0121] The present disclosure may be a system, a method and / or a computer program product. The computer program product may include a computer-readable storage medium carrying computer-readable program instructions for causing a processor to implement various aspects of the present disclosure.

[0122] A computer-readable storage medium may be a tangible device that can hold and store instructions used by an instruction execution device. A computer-readable storage medium may be, for example, but not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing. More specific examples of computer-readable storage media (a non-exhaustive list) include: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a static random access memory (SRAM), a portable compact disk read-only memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, a mechanical encoding device, such as a punch card or a raised structure in a groove on which instructions are stored, and any suitable combination of the foregoing. As used herein, a computer-readable storage medium is not to be interpreted as a transient signal per se, such as a radio wave or other freely propagating electromagnetic wave, an electromagnetic wave propagating through a waveguide or other transmission medium (e.g., a light pulse through a fiber optic cable), or an electrical signal transmitted through a wire.

[0123] The computer-readable program instructions described herein can be downloaded from a computer-readable storage medium to each computing / processing device, or downloaded to an external computer or external storage device via a network, such as the Internet, a local area network, a wide area network, and / or a wireless network. The network can include copper transmission cables, optical fiber transmissions, wireless transmissions, routers, firewalls, switches, gateway computers, and / or edge servers. The network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards the computer-readable program instructions for storage in the computer-readable storage medium in each computing / processing device.

[0124] The computer program instructions for performing the operation of the present disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-related instructions, microcode, firmware instructions, state setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages, such as Smalltalk, C++, etc., and conventional procedural programming languages, such as "C" language or similar programming languages. Computer-readable program instructions may be executed completely on a user's computer, partially on a user's computer, as an independent software package, partially on a user's computer, partially on a remote computer, or completely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., using an Internet service provider to connect via the Internet). In some embodiments, an electronic circuit, such as a programmable logic circuit, a field programmable gate array (FPGA), or a programmable logic array (PLA), may be customized by utilizing the state information of the computer-readable program instructions, and the electronic circuit may execute the computer-readable program instructions, thereby realizing various aspects of the present disclosure.

[0125] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents, and any modifications or equivalent replacements that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A battery vehicle load monitoring method based on EMD feature extraction and SVM, characterized in that: The following steps are involved: Step 1, collecting the user's historical current original signal data; Step 2: determine whether there is a transient device turn-on based on the current data, and if so, extract the turn-on event data based on the sliding window; Step 3, extracting current wavelet features through EMD transformation based on the start-up event data and constructing a load feature data set; Step 4: Use a support vector machine to identify the load feature data set to determine whether a battery vehicle power load is generated.

2. The battery vehicle load monitoring method based on EMD feature extraction and SVM according to claim 1 is characterized in that: In step 1, the user's historical current original signal is collected through the user's intelligent terminal, where the terminal sampling frequency is 6.4khz.

3. The battery vehicle load monitoring method based on EMD feature extraction and SVM according to claim 1 is characterized in that: Step 2 includes: extracting load current waveform data from the original current signal data, judging whether a transient start-up event of the equipment occurs by the extreme value change of the load current waveform data, when the extreme value change is greater than the threshold, an event occurs, otherwise no event occurs; based on the mutation principle, a sliding window is used to extract the data in the window as the start-up event data.

4. The battery vehicle load monitoring method based on EMD feature extraction and SVM according to claim 3 is characterized in that: Step 2 specifically includes: Step 2.1, extracting load current waveform data I based on historical current original signal data; Step 2.2, determine whether the extreme value of the load current waveform data has a sudden change, if so, execute step 2.3, where the mutation principle is: take w as a sliding window, if the difference between the current waveform data in the two windows is greater than the set threshold, an electrical equipment start event occurs; Step 2.3: extract the data in the sliding window as the power-consuming device startup event data.

5. The battery vehicle load monitoring method based on EMD feature extraction and SVM according to claim 4 is characterized in that: The two windows are respectively I s with I s+1 , the mutation principle formula is as follows: I s =[i1,i2,…,i w ] I s,α =sort(I s )(0.75·n) I s+1,α -I s,α >TH In the formula, i w represents the wth data in a sliding window of size w, sort is a function for arranging from small to large, I s,α is the third quartile, and TH is the event mutation threshold.

6. The battery vehicle load monitoring method based on EMD feature extraction and SVM according to claim 1 is characterized in that: Step 3 includes decomposing the voltage original signal and the current original signal corresponding to the power-on event data of the electrical equipment, obtaining a multi-level decomposition to generate an intrinsic mode function, calculating the intrinsic mode function energy and the intrinsic mode function steepness of the intrinsic mode function of different levels respectively, and establishing a feature data set.

7. The battery vehicle load monitoring method based on EMD feature extraction and SVM according to claim 6 is characterized in that: Step 3 specifically includes: Step 3.1, respectively decompose the voltage original signal and current original signal corresponding to the power-on event data of the power-consuming equipment extracted in step 2, and obtain the intrinsic mode function S generated by multi-level decomposition I : Where n is the number of intrinsic mode functions obtained by decomposition, c j is the jth intrinsic mode function, r is the residual signal; Step 3.2, respectively calculate the intrinsic mode function energy and intrinsic mode function steepness of the intrinsic mode functions of different levels and establish a characteristic data set.

8. The battery vehicle load monitoring method based on EMD feature extraction and SVM according to claim 7 is characterized in that: The intrinsic mode function energy calculation formula is: In the formula, N is the number of data points contained in the sample, c j is the jth eigenmode function, c j (i) is the i-th data point of the j-th eigenmode function; Steepness of the intrinsic mode function: In the formula, mean is the average value, μ is j is the mean value of the jth effective intrinsic mode function, σ j is the standard deviation of the jth effective eigenmode function.

9. The battery vehicle load monitoring method based on EMD feature extraction and SVM according to claim 1 is characterized in that: Step 4 includes: dividing the data feature set into a training set and a test set, taking the battery vehicle start-up event as the positive sample X1 and the other electrical equipment start-up events as the negative sample X2, constructing a hyperplane to identify the load data, and when the result is in the positive sample interval of the hyperplane, it is represented as the battery vehicle being turned on, and when it is in the negative sample interval, it is represented as other electrical equipment.

10. The battery vehicle load monitoring method based on EMD feature extraction and SVM according to claim 9 is characterized in that: Step 4.1: Divide the data feature set described in step 3 into a training set Ω Train and the test set Ω Test , the training set Ω Train The battery vehicle start-up event is taken as the positive sample X1, and the other electrical equipment start-up events are taken as the negative sample X2; Step 4.2, construct a hyperplane H:w·x+b=0 to perform regression analysis on the data of different power-consuming equipment startup events, so that the two types of data X1 and X2 in the data set are separated, where w is the normal vector of the hyperplane and b represents the classification threshold, specifically: In the formula, ε i is the penalty factor; ξ is the penalty parameter used to penalize misclassification; x i is the eigenvalue of the i-th sample; n1 is the number of samples X1, n2 is the number of samples X2; min f(w,ε) is the minimum value of the target optimization function; Step 4.3, based on the test set Ω Test Optimize the normal vector and classification threshold of the hyperplane H to obtain the optimal plane model, input the data feature set obtained in real time into the optimal plane model, and when the result is in the positive sample interval of the hyperplane, it is represented as an electric vehicle, and it is judged that the electric vehicle power load is generated in the power load data of the user; when it is in the negative sample interval, it is represented as other electrical equipment.

11. The battery vehicle load monitoring system based on EMD feature extraction and SVM is characterized by: The system comprises: a data acquisition module for acquiring user voltage original signal data and current original signal data; An event extraction module, used to determine the transient start of the device based on the current data, and extract the start event based on the sliding window; A feature extraction module is used to extract voltage and current wavelet features based on the start-up event data through a fast wavelet transform algorithm to obtain a data feature set; The result judgment module is used to judge the battery vehicle event through the support vector machine based on the data feature set.

12. A terminal comprising a processor and a storage medium; characterized in that: The storage medium is used to store instructions; The processor is configured to operate according to the instructions to execute the steps of the method according to any one of claims 1-9.

13. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the method according to any one of claims 1 to 9 are implemented.

Images