Load identification method, system and equipment based on power time sequence characteristics and medium

Abstract

The invention relates to a load identification method, system and device based on power time sequence characteristics and a medium, and belongs to the technical field of load monitoring, and the method comprises the steps: obtaining a time sequence generated based on the active power of a cycle; performing feature extraction on the time sequence to obtain statistical features; performing phase-space reconstruction on the time sequence, and calculating topological characteristics of the phase space of the time sequence; splicing the statistical features and the topological features of the phase space to obtain spliced features; and inputting the splicing features into a machine learning classifier to obtain a final load identification result. Aiming at non-intrusive load monitoring NILM application in the field of electrical and intelligent power grids, in feature extraction, in order to improve the accuracy of active power in equipment classification and identification, TDA is used as a feature extraction method, a large number of nonlinear shape features are mined from the active power to identify loads, and the accuracy of the active power in equipment classification and identification is improved. The method has superiority in classification identification accuracy and calculation time.

Description

Technical Field

[0001] The present disclosure belongs to the technical field of load monitoring, and in particular, relates to a load identification method, system, device and medium based on power time series characteristics. Background Art

[0002] With the development of smart grids, it is increasingly important to analyze consumers' electricity usage patterns and provide energy-saving guidance. A relatively simple and direct way is to set up an intrusive device that can sample every few seconds or more to regularly determine the total household electricity consumption. However, in actual scenarios, monitoring electricity consumption in this way is neither convenient nor costly. Against this background, the concept of non-intrusive load monitoring (NILM), also known as energy decomposition, came into being. NILM aims to use the total load consumption of the household to infer the usage of specific appliances without the need for additional sensors. Therefore, NILM provides a cost-effective solution for energy consumption monitoring in actual scenarios.

[0003] For the NILM method based on load events, the key is feature extraction. Many studies have focused on identifying traditional physical characteristics of electrical appliances, such as current, power, or various statistical properties derived from measurements. However, the accuracy of classification and recognition achieved has not yet reached the desired level. The traditional method of extracting one-dimensional features based on physical quantities has a fast operation speed, but these features are usually not stable enough, because a small perturbation of the original sequence will bring about obvious changes in some features. Therefore, it is necessary to introduce some features with more stable properties.

[0004] Topological data analysis (TDA) is a relatively new branch of data analysis that uses techniques from topology to analyze data. TDA has been successfully applied in some fields, such as biomolecular chemistry, drug design, image analysis, time series data analysis, and network analysis. In particular, the winners of the D3R Challenge have incorporated TDA into their algorithmic methods. In addition, TDA can be technically synergistically integrated with some machine learning techniques, including deep learning and statistical methods. With the continuous updating of technology, some deep learning algorithms have also been used to improve traditional feature extraction methods, but the features extracted by these methods do not have clear mathematical meanings and take a long time to calculate.

[0005] Therefore, it is necessary to provide a new load identification method, system, device and medium based on power time series characteristics to solve the above technical problems. Summary of the invention

[0006] The purpose of the present disclosure is to provide a load identification method, system, device and medium based on power time series characteristics in order to solve the above problems.

[0007] The present invention achieves the above-mentioned purpose through the following technical solutions: A load identification method based on power time series characteristics comprises the following steps: Get the time series generated based on one cycle active power; Extracting features from the time series to obtain statistical features; Reconstructing the phase space of the time series and calculating the topological characteristics of the phase space; Splicing the statistical feature with the topological feature of the phase space to obtain a spliced ​​feature; The splicing features are input into a machine learning classifier to obtain a final load identification result.

[0008] As a further optimization scheme of the present disclosure, the statistical features include absolute energy, mean value, standard deviation, skewness, kurtosis, absolute average change, average second-order difference center and median.

[0009] As a further optimization solution of the present disclosure, feature extraction is performed on the time series to obtain statistical features including: set up is the cycle active power sequence; The absolute energy is the sum of the squares of the components of the time series, and the calculation formula is as follows: ; The average value is the mean of the time series, and the calculation formula is as follows: ; The standard deviation is calculated as follows: ; The skewness is the third-order standardized moment of the time series, and the calculation formula is as follows: ; The kurtosis is the fourth-order standardized moment of the time series, and the calculation formula is as follows: ; The absolute average change is the mean of the absolute values ​​of the continuous change values ​​of the time series, and the calculation formula is as follows: ; The average second-order difference center is the mean of the second-order change of the time series, and the calculation formula is as follows: ; The median is the median of each value in the time series arranged from small to large.

[0010] As a further optimization scheme of the present disclosure, the phase space reconstruction of the time series includes: The one-dimensional time series is used Indicates that, given the delay parameter , the Takens phase space reconstruction point corresponding to time t is defined as: ; Where d is the embedding dimension, ; Given parameters d and ,Phase space reconstruction converts time series into point cloud data.

[0011] As a further optimization solution of the present disclosure, the topological characteristics include Betti number, amplitude and persistence entropy.

[0012] As a further optimization scheme of the present disclosure, the topological characteristics of the phase space are calculated including: Based on the phase space reconstruction, the point cloud data is filtered to obtain a filtered complex; Obtaining a persistence graph based on the filter complex; The persistence diagram is used Indicates that the Betti number, amplitude and persistence entropy are calculated; the Betti number is the number of points in the persistence graph D in the sense of multiplicity; the amplitude is the farthest distance from the midpoint of the persistence graph D to the diagonal; the persistence entropy is the Shannon entropy of the persistence graph D, and the formula is as follows: ; ; in, is the horizontal coordinate of the ith point in the persistence diagram, is the ordinate of the i-th point in the persistence graph.

[0013] As a further optimization solution of the present disclosure, the machine learning classifier adopts XGBoost classifier.

[0014] A load identification system based on power time series characteristics, comprising: A data acquisition module, used for acquiring a time series generated based on a cycle active power; A statistical feature extraction module, used to extract features from the time series to obtain statistical features; A topological feature calculation module, used for reconstructing the phase space of the time series and calculating the topological features of its phase space; A feature splicing module, used for splicing the statistical feature and the topological feature of the phase space to obtain a splicing feature; The load identification module is used to input the splicing features into a machine learning classifier to obtain a final load identification result.

[0015] As a further optimization scheme of the present disclosure, the statistical features include absolute energy, mean value, standard deviation, skewness, kurtosis, absolute average change, average second-order difference center and median.

[0016] As a further optimization solution of the present disclosure, the statistical feature extraction module extracts features from the time series to obtain statistical features including: set up is the cycle active power sequence; The absolute energy is the sum of the squares of the components of the time series, and the calculation formula is as follows: ; The average value is the mean of the time series, and the calculation formula is as follows: ; The standard deviation is calculated as follows: ; The skewness is the third-order standardized moment of the time series, and the calculation formula is as follows: ; The kurtosis is the fourth-order standardized moment of the time series, and the calculation formula is as follows: ; The absolute average change is the mean of the absolute values ​​of the continuous change values ​​of the time series, and the calculation formula is as follows: ; The average second-order difference center is the mean of the second-order change of the time series, and the calculation formula is as follows: ; The median is the median of each value in the time series arranged from small to large.

[0017] As a further optimization solution of the present disclosure, the topological feature calculation module performs phase space reconstruction on the time series including: The one-dimensional time series is used Indicates that, given the delay parameter , the Takens phase space reconstruction point corresponding to time t is defined as: ; Where d is the embedding dimension, ; Given parameters d and ,Phase space reconstruction converts time series into point cloud data.

[0018] As a further optimization solution of the present disclosure, the topological characteristics include Betti number, amplitude and persistence entropy.

[0019] As a further optimization solution of the present disclosure, the topological feature calculation module calculates the topological features of the phase space including: Based on the phase space reconstruction, the point cloud data is filtered to obtain a filtered complex; Obtaining a persistence graph based on the filter complex; The persistence diagram is used Indicates that the Betti number, amplitude and persistence entropy are calculated; the Betti number is the number of points in the persistence graph D in the sense of multiplicity; the amplitude is the farthest distance from the midpoint of the persistence graph D to the diagonal; the persistence entropy is the Shannon entropy of the persistence graph D, and the formula is as follows: ; ; in, is the horizontal coordinate of the ith point in the persistence diagram, is the ordinate of the i-th point in the persistence graph.

[0020] As a further optimization solution of the present disclosure, the machine learning classifier adopts XGBoost classifier.

[0021] An electronic device comprises a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other via the communication bus; Memory, for storing computer programs; The processor is used to implement a load identification method based on power time series characteristics when executing a program stored in the memory.

[0022] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements a load identification method based on power time series characteristics.

[0023] The beneficial effects of the present disclosure are: The present invention discloses non-intrusive load monitoring (NILM) applications in the electrical and smart grid fields. In the feature extraction step, in order to improve the accuracy of active power in equipment classification and identification, TDA is used as a feature extraction method to mine a large number of nonlinear shape features from the active power to identify the load. The comparison with the experimental results of other methods proves the superiority of this method in classification and identification accuracy and calculation time. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 is a flow chart of a method in an embodiment of the present disclosure; Figure 2 is a system structure block diagram in an embodiment of the present disclosure; Figure 3 It is a block diagram of the device structure in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0025] The present application is further described in detail below in conjunction with the accompanying drawings. It is necessary to point out here that the following specific implementation methods are only used to further illustrate the present application and cannot be understood as limiting the scope of protection of the present application. Technical personnel in this field can make some non-essential improvements and adjustments to the present application based on the above application content.

[0026] like Figure 1 As shown, a load identification method based on power time series characteristics includes the following steps: Get the time series generated based on one cycle active power; Extracting features from the time series to obtain statistical features; Reconstructing the phase space of the time series and calculating the topological characteristics of the phase space; Splicing the statistical feature with the topological feature of the phase space to obtain a spliced ​​feature; The splicing features are input into a machine learning classifier to obtain a final load identification result.

[0027] Among them, the statistical features mainly include absolute energy, mean value, standard deviation, skewness, kurtosis, absolute average change, average second-order difference center and median, etc. The topological features mainly include Betti number, amplitude and persistence entropy, etc.

[0028] Feature extraction is performed on the time series to obtain statistical features including: set up is the cycle active power sequence; The absolute energy is the sum of the squares of the components of the time series, and the calculation formula is as follows: ; The average value is the mean of the time series, and the calculation formula is as follows: ; The standard deviation is calculated as follows: ; The skewness is the third-order standardized moment of the time series, which is a measure of the skewness direction and degree of the statistical data distribution. The calculation formula is as follows: ; The kurtosis is the fourth-order standardized moment of the time series, which is a characteristic quantity describing the peak height of the probability density distribution curve at the mean value. The calculation formula is as follows: ; The absolute average change is the mean of the absolute values ​​of the continuous change values ​​of the time series, and the calculation formula is as follows: ; The average second-order difference center is the mean of the second-order change of the time series, and the calculation formula is as follows: ; The median is the median of each value in the time series arranged from small to large.

[0029] The time series is reconstructed in phase space, and the topological features of the phase space are calculated, including: In order to analyze the topological properties of the time series, it is necessary to first convert the time series into a point cloud representation. The method used in this disclosure is phase space reconstruction, and its theoretical basis comes from the following theorem: (Takens embedding theorem) If For one dimensional compact manifold, and For an almost everywhere smooth mapping, given the delay parameter ,for Points in ,definition ,but is the embedding mapping between manifolds.

[0030] Specifically for the scenario to be studied in this disclosure, the one-dimensional time series is used Indicates that, given the delay parameter , the Takens phase space reconstruction point corresponding to time t is defined as: ; Where d is the embedding dimension, ; Given parameters d and , phase space reconstruction converts time series into point cloud data; Based on the phase space reconstruction, Vietoris-Rips filtering is used to filter the point cloud data to obtain the filtered complex; Based on the filter complex, a persistence graph is obtained, where the persistence graph is a 2D graphical representation of the topological features of the filter complex and is composed of a finite number of points, each of which corresponds to a topological generator of the filter complex; The persistence diagram is used Indicates that the Betti number, amplitude and persistence entropy are calculated; the Betti number is the number of points in the persistence graph D in the sense of multiplicity; the amplitude is the farthest distance between the midpoint of the persistence graph D and the diagonal line, and the implementation of the present disclosure uses several different distances when calculating the amplitude; the persistence entropy is the Shannon entropy of the persistence graph D, and the formula is as follows: ; ; in, is the horizontal coordinate of the ith point in the persistence diagram, is the ordinate of the i-th point in the persistence graph.

[0031] In this embodiment, a public dataset PLAID is selected for experiments, wherein 20% of the data is randomly selected as a test set, and 80% is used as a training set, and is validated by 5-fold cross validation.

[0032] PLAID is a high-frequency dataset containing the voltage and current of a single load, with a sampling frequency of 30 kHz. It includes 11 categories of common household appliances, 312 different appliances, and 1793 samples.

[0033] In the experiments of the present disclosure, XGBoost is selected as the machine learning classifier of the model proposed in the present disclosure. The XGBoost classifier is a powerful, flexible and efficient machine learning tool. It provides excellent performance and a wide range of functions through the gradient boosting framework and is suitable for solving various complex classification problems. Therefore, XGBoost is selected as the machine learning classifier.

[0034] As a comparison item, we choose one-dimensional convolutional neural network for comparison. One-dimensional convolutional neural network (1D CNN) is a special type of convolutional neural network, which is mainly used to process sequence data, such as audio signals, time series data, or any other type of one-dimensional sequence. Unlike traditional two-dimensional convolutional neural network (2D CNN), which is usually used to process image data, 1D CNN uses one-dimensional convolution kernel to extract features in the sequence, which is particularly suitable for processing one-dimensional sequence data. Through its combination of convolution and pooling layers, 1D CNN can effectively extract and utilize temporal dependencies and patterns in sequence data, providing a powerful solution for various time series analysis tasks.

[0035] In addition, the new method proposed in the present disclosure is also compared with some methods based on voltage and current binary images. Graph signal processing is a method for processing and analyzing signals defined on a graph structure. In the context of non-intrusive load monitoring, graph signal processing can be used to extract and analyze the characteristics of the load for identification.

[0036] The 2D convolutional neural network is a commonly used method for processing image data in recent years. It extracts features through convolution kernels through convolution layers and pooling layers.

[0037] Performance Use of Load Identification Methods Scores are evaluated. The score is defined as ,in For accuracy, is the recall rate. It is a more meaningful classification evaluation indicator that takes into account class imbalance.

[0038] The final experimental results are shown in Table 1:

[0039] Table 1 Identification of electrical loads value It is not difficult to see from the above results that the method proposed in the present disclosure has a shorter calculation time than graph signal processing and convolutional neural network, and has reached above 0.9 in the identification of all electrical loads except refrigerators and washing machines.

[0040] In particular, for complex multi-mode and multi-frequency electrical appliances such as refrigerators and washing machines, the method proposed in the present disclosure has a significant improvement in recognition effect compared to the method based on voltage and current binary images. For the three devices of fans / heaters / incandescent lamps, the method proposed in the present disclosure has a significant improvement in recognition effect compared to the one-dimensional convolutional neural network.

[0041] like Figure 2 As shown, an embodiment of the present disclosure provides a load identification system based on power time series characteristics, including: A data acquisition module, used for acquiring a time series generated based on a cycle active power; A statistical feature extraction module, used to extract features from the time series to obtain statistical features; A topological feature calculation module, used for reconstructing the phase space of the time series and calculating the topological features of its phase space; A feature splicing module, used for splicing the statistical feature and the topological feature of the phase space to obtain a splicing feature; The load identification module is used to input the splicing features into a machine learning classifier to obtain a final load identification result.

[0042] The implementation process of the functions and effects of each module in the above system is specifically described in the implementation process of the corresponding steps in the above method, which will not be repeated here.

[0043] For the system embodiment, since it basically corresponds to the method embodiment, the relevant parts can refer to the partial description of the method embodiment. The system embodiment described above is only schematic, wherein the modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical modules, that is, they may be located in one place, or they may be distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the disclosed solution. A person of ordinary skill in the art can understand and implement it without paying any creative work.

[0044] In the above embodiments, any number of all modules can be combined into one module for implementation, or any one of the modules can be split into multiple modules. Alternatively, at least part of the functions of one or more of these modules can be combined with at least part of the functions of other modules and implemented in one module. At least one of all modules can be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on a chip, a system on a substrate, a system on a package, an application specific integrated circuit (ASIC), or can be implemented by hardware or firmware such as any other reasonable way of integrating or packaging the circuit, or implemented in any one of the three implementation methods of software, hardware and firmware or in a suitable combination of any of them. Alternatively, at least one of all modules can be at least partially implemented as a computer program module, and when the computer program module is run, the corresponding function can be executed.

[0045] See also Figure 3 , an electronic device provided by an embodiment of the present disclosure includes a processor 1110, a communication interface 1120, a memory 1130 and a communication bus 1140, wherein the processor 1110, the communication interface 1120, and the memory 1130 communicate with each other through the communication bus 1140; Memory 1130, for storing computer programs; The processor 1110 is used to implement the load identification method based on power time series characteristics as shown below when executing the program stored in the memory 1130.

[0046] The communication bus 1140 may be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The communication bus 1140 may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, only one thick line is used in the figure, but it does not mean that there is only one bus or one type of bus.

[0047] The communication interface 1120 is used for communication between the above electronic device and other devices.

[0048] The memory 1130 may include a random access memory (RAM) or a non-volatile memory, such as at least one disk memory. Optionally, the memory 1130 may also be at least one storage device located away from the processor 1110.

[0049] The above-mentioned processor 1110 can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components.

[0050] The embodiment of the present disclosure further provides a computer-readable storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the load identification method based on power time series characteristics as described above is implemented.

[0051] The computer-readable storage medium may be included in the device / apparatus described in the above embodiment; or it may exist independently without being assembled into the device / apparatus. The above computer-readable storage medium carries one or more programs. When the above one or more programs are executed, the load identification method based on power time series characteristics according to the embodiment of the present disclosure is implemented.

[0052] According to an embodiment of the present disclosure, a computer-readable storage medium may be a non-volatile computer-readable storage medium, for example, may include but is not limited to: 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 portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, apparatus, or device.

[0053] The above-mentioned embodiments only express several implementation methods of the present disclosure, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present disclosure. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present disclosure, and these all belong to the protection scope of the present disclosure.

Claims

1. A load identification method based on power time series characteristics, characterized in that: The following steps are involved: Get the time series generated based on one cycle active power; Extracting features from the time series to obtain statistical features; Reconstructing the phase space of the time series and calculating the topological characteristics of the phase space; Splicing the statistical feature with the topological feature of the phase space to obtain a spliced ​​feature; The splicing features are input into a machine learning classifier to obtain a final load identification result.

2. A load identification method based on power time series characteristics according to claim 1, characterized in that: The statistical characteristics include absolute energy, mean, standard deviation, skewness, kurtosis, absolute mean variation, mean second-order difference center and median.

3. A load identification method based on power time series characteristics according to claim 2, characterized in that: Feature extraction is performed on the time series to obtain statistical features including: set up is the cycle active power sequence; The absolute energy is the sum of the squares of the components of the time series, and the calculation formula is as follows: ； The average value is the mean of the time series, and the calculation formula is as follows: ； The standard deviation is calculated as follows: ； The skewness is the third-order standardized moment of the time series, and the calculation formula is as follows: ； The kurtosis is the fourth-order standardized moment of the time series, and the calculation formula is as follows: ； The absolute average change is the mean of the absolute values ​​of the continuous change values ​​of the time series, and the calculation formula is as follows: ； The average second-order difference center is the mean of the second-order change of the time series, and the calculation formula is as follows: ； The median is the median of each value in the time series arranged from small to large.

4. The load identification method based on power time series characteristics according to claim 1 is characterized in that: Reconstructing the phase space of the time series includes: The one-dimensional time series is used Indicates that, given the delay parameter , the Takens phase space reconstruction point corresponding to time t is defined as: ； Where d is the embedding dimension, ; Given parameters d and ,Phase space reconstruction converts time series into point cloud data.

5. The load identification method based on power time series characteristics according to claim 1 is characterized in that: The topological characteristics include Betti number, amplitude and persistence entropy.

6. A load identification method based on power time series characteristics according to claim 4, characterized in that: The topological features of the computational phase space include: Based on the phase space reconstruction, the point cloud data is filtered to obtain a filtered complex; Obtaining a persistence graph based on the filter complex; The persistence diagram is used Indicates that the Betti number, amplitude and persistence entropy are calculated; the Betti number is the number of points in the persistence graph D in the sense of multiplicity; the amplitude is the farthest distance from the midpoint of the persistence graph D to the diagonal; the persistence entropy is the Shannon entropy of the persistence graph D, and the formula is as follows: ； ； in, is the horizontal coordinate of the ith point in the persistence diagram, is the ordinate of the i-th point in the persistence graph.

7. The load identification method based on power time series characteristics according to claim 1 is characterized in that: The machine learning classifier adopts XGBoost classifier.

8. A load identification system based on power time series characteristics, characterized in that: include: A data acquisition module, used for acquiring a time series generated based on a cycle active power; A statistical feature extraction module, used to extract features from the time series to obtain statistical features; A topological feature calculation module, used for reconstructing the phase space of the time series and calculating the topological features of its phase space; A feature splicing module, used for splicing the statistical feature and the topological feature of the phase space to obtain a splicing feature; The load identification module is used to input the splicing features into a machine learning classifier to obtain a final load identification result.

9. A load identification system based on power time series characteristics according to claim 8, characterized in that: The statistical characteristics include absolute energy, mean, standard deviation, skewness, kurtosis, absolute mean variation, mean second-order difference center and median.

10. A load identification system based on power time series characteristics according to claim 9, characterized in that: The statistical feature extraction module extracts features from the time series to obtain statistical features including: set up is the cycle active power sequence; The absolute energy is the sum of the squares of the components of the time series, and the calculation formula is as follows: ； The average value is the mean of the time series, and the calculation formula is as follows: ； The standard deviation is calculated as follows: ； The skewness is the third-order standardized moment of the time series, and the calculation formula is as follows: ； The kurtosis is the fourth-order standardized moment of the time series, and the calculation formula is as follows: ； The absolute average change is the mean of the absolute values ​​of the continuous change values ​​of the time series, and the calculation formula is as follows: ； The average second-order difference center is the mean of the second-order change of the time series, and the calculation formula is as follows: ； The median is the median of each value in the time series arranged from small to large.

11. A load identification system based on power time series characteristics according to claim 8, characterized in that: The topological feature calculation module performs phase space reconstruction on the time series, including: The one-dimensional time series is used Indicates that, given the delay parameter , the Takens phase space reconstruction point corresponding to time t is defined as: ； Where d is the embedding dimension, ; Given parameters d and ,Phase space reconstruction converts time series into point cloud data.

12. A load identification system based on power time series characteristics according to claim 8, characterized in that: The topological characteristics include Betti number, amplitude and persistence entropy.

13. A load identification system based on power time series characteristics according to claim 11, characterized in that: The topological feature calculation module calculates the topological features of the phase space including: Based on the phase space reconstruction, the point cloud data is filtered to obtain a filtered complex; Obtaining a persistence graph based on the filter complex; The persistence diagram is used Indicates that the Betti number, amplitude and persistence entropy are calculated; the Betti number is the number of points in the persistence graph D in the sense of multiplicity; the amplitude is the farthest distance from the midpoint of the persistence graph D to the diagonal; the persistence entropy is the Shannon entropy of the persistence graph D, and the formula is as follows: ； ； in, is the horizontal coordinate of the ith point in the persistence diagram, is the ordinate of the i-th point in the persistence graph.

14. A load identification system based on power time series characteristics according to claim 8, characterized in that: The machine learning classifier adopts XGBoost classifier.

15. An electronic device, characterized in that: It includes a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other through the communication bus; Memory, for storing computer programs; The processor is used to implement the load identification method based on power time series characteristics described in any one of claims 1 to 7 when executing the program stored in the memory.

16. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the load identification method based on power time series characteristics described in any one of claims 1 to 7 is implemented.

Images