Electric bicycle charging monitoring method, system and device and storage medium

Abstract

The invention discloses an electric bicycle charging monitoring method, system and device and a storage medium, and belongs to the field of electric bicycle charging monitoring, and the method comprises the following steps: obtaining a residential electricity load in real time; performing wavelet packet decomposition on the residential electricity load to obtain a transient load of the high-frequency signal, and positioning a load electricity utilization event based on the transient load of the high-frequency signal; obtaining steady-state load data of a plurality of periods before and after the time point of the load power utilization event, calculating a single load sample based on the steady-state load data, and extracting a feature set of the single load sample; and clustering the feature set of the single load sample, inputting the clustered feature set into a LightGBM decision tree classifier which is trained in advance, and outputting an electric bicycle charging monitoring result by the LightGBM decision tree classifier. The number of features required by the monitoring method can be obviously reduced, and the applicability and the identification precision of the method are improved.

Description

Technical Field

[0001] The invention belongs to the field of electric bicycle charging monitoring, and relates to an electric bicycle charging monitoring method, system, device and storage medium. Background Art

[0002] Fire accidents caused by electric bicycles charging indoors occur increasingly frequently, which has become a problem attracting much attention in the field of public safety. The batteries of electric bicycles are mainly lithium batteries and lead-acid batteries, and these batteries may cause fires during the charging process due to overcharging, overheating, electrical faults and other reasons. In recent years, the proportion of fires caused by electric bicycles charging indoors in the total number of fires has been increasing year by year, and 90% of electric bicycle fires are caused by improper charging. These accidents have caused huge economic losses and directly threatened the lives and safety of residents. Especially in communities with dense residential buildings, electric bicycles charging indoors are likely to cause the rapid spread of fires, making rescue difficult and the consequences extremely serious. Even if it is prohibited to charge electric bicycles indoors and the use of centralized charging piles is encouraged, the actual implementation effect is not ideal. Due to concerns about the loss of electric vehicles, insufficient centralized charging facilities or costs, residents still choose to push electric bicycles into their homes for charging. This kind of illegal behavior makes the existing management means unable to effectively avoid potential safety hazards, and there is an urgent need for technical means to support the monitoring and management of indoor charging behavior.

[0003] Non-intrusive load monitoring technology is a technology that analyzes the total power signal of power users to identify the types and operating states of electrical equipment they use. Compared with the traditional intrusive monitoring method, non-intrusive load monitoring technology does not require installing sensors on each electrical equipment. Only through single-point power data collection and using signal processing and machine learning algorithms, equipment identification and status monitoring can be completed. This method has low cost and convenient deployment, so it has received extensive attention in the fields of smart grid, energy management, user behavior analysis, etc. At present, non-intrusive load monitoring technology mainly extracts electrical characteristics (voltage, current, power waveform, etc.) of different loads through methods such as wavelet transform and wavelet packet decomposition, and combines data-driven methods (k-means clustering, neural network, support vector machine, Bayesian classification, deep learning, etc.) to achieve load identification. However, there are still problems in the current non-intrusive monitoring of electric bicycles charging indoors, such as insufficient robustness of load feature extraction, excessive number of model features, and relatively high computational complexity of real-time monitoring. Summary of the Invention

[0004] The purpose of the invention is to overcome the above-mentioned shortcomings of the prior art and provide an electric bicycle charging monitoring method, system, device and storage medium, which can significantly reduce the number of features required by the monitoring method and improve the applicability and identification accuracy of the method.

[0005] To achieve the above purpose, the invention adopts the following technical solutions: An electric bicycle charging monitoring method, including the following processes: Obtain the residential electricity load in real time; Perform wavelet packet decomposition on the residential electricity load to obtain the transient load of the high-frequency signal, and locate the load power consumption event based on the transient load of the high-frequency signal; Obtain the steady-state load data of several cycles before and after the time point of the load power consumption event, calculate a single load sample based on the steady-state load data, and extract the feature set of the single load sample; After clustering the feature set of the single load sample, input it into a pre-trained LightGBM decision tree classifier, and the LightGBM decision tree classifier outputs the electric bicycle charging monitoring result.

[0006] Preferably, the specific process of performing wavelet packet decomposition on the residential electricity load to obtain the transient load of the high-frequency signal and locating the load power consumption event based on the transient load of the high-frequency signal is as follows: Perform multi-layer wavelet packet decomposition on the load current in the residential electricity load to obtain the transient load of the high-frequency signal. If the transient load of the high-frequency signal changes suddenly, a load power consumption event occurs.

[0007] Preferably, the specific process of calculating a single load sample based on the steady-state load data and extracting the feature set of the single load sample is: calculate the single load sample of the steady-state load data of several cycles according to the superposition principle, and extract the feature set of the single load sample through Fourier transform and wavelet packet decomposition.

[0008] Preferably, the feature set of the single load sample includes: instantaneous power, active power, THD, power factor, effective current value, current peak value, DC component of current, fundamental component, 3rd harmonic component, 5th harmonic component, 7th harmonic component, and the energy of different frequency bands decomposed by wavelet packet 2 layers.

[0009] Preferably, the training process of the LightGBM decision tree classifier is: use the feature set of the single load sample as the training set to train the LightGBM decision tree classifier, and set the maximum number of leaf nodes, convergence accuracy, and number of iterations; optimize the LightGBM decision tree classifier by minimizing the loss function to obtain the trained LightGBM decision tree classifier.

[0010] Preferably, the specific process of clustering the feature set of the single load sample is: screen the feature set of the single load sample through FCM clustering, extract the dominant features, and use the dominant features as the training set.

[0011] Preferably, the specific process of screening the feature set of the single load sample through FCM clustering and extracting the dominant features is: Calculate the clustering accuracy of different feature combinations in the feature set of a single load sample; select the feature combination with the highest clustering accuracy as the dominant feature.

[0012] An electric bicycle charging monitoring system, comprising: A load acquisition module for acquiring the residential electricity load in real time; A load power consumption event positioning module for performing wavelet packet decomposition on the residential electricity load to obtain the transient load of the high-frequency signal, and positioning the load power consumption event based on the transient load of the high-frequency signal; A feature extraction module for obtaining the steady-state load data of several cycles before and after the time point of the load power consumption event, calculating a single load sample based on the steady-state load data, and extracting the feature set of the single load sample; A result output module for clustering the feature set of the single load sample and inputting it into a pre-trained LightGBM decision tree classifier, and the LightGBM decision tree classifier outputs the electric bicycle charging monitoring result.

[0013] A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, and when the processor executes the computer program, the steps of the electric bicycle charging monitoring method are implemented.

[0014] A computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, the steps of the electric bicycle charging monitoring method are implemented.

[0015] Compared with the prior art, the present invention has the following beneficial effects: The present invention adopts a LightGBM decision tree classifier, which not only reduces the requirement for the number of features, but also has strong robustness, thereby reducing the computational complexity, and can adapt to the changes of different users and different electric bicycle charging loads, ensuring the wide applicability of the system. Through a non-intrusive monitoring method, direct intervention in the user's power equipment is avoided, and real-time monitoring of current and voltage is adopted to ensure the accuracy and security of the monitoring data. This monitoring method can identify the illegal behavior of electric bicycles charging indoors in real time and provide effective monitoring data to help reduce the risk of fire accidents caused by illegal charging and improve the safety guarantee of residents. Compared with the traditional electric bicycle charging monitoring technology, the present invention has strong adaptability and promotion potential. Its low feature requirement and high generality enable this method to adapt to the monitoring requirements of different regions and different types of electric bicycle loads, and can also be effectively applied in other types of load monitoring scenarios, facilitating rapid promotion. Description of the Drawings

[0016] Figure 1Flowchart of the electric bicycle charging monitoring method according to Embodiment 1 of the present invention; Figure 2 Flowchart of the electric bicycle charging monitoring method according to Embodiment 2 of the present invention; Figure 3 Schematic diagram of wavelet packet decomposition of the electric bicycle according to the present invention; Figure 4 Schematic diagram of the FCM clustering result of the dominant features of the present invention. Detailed implementation manners

[0017] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application.

[0018] Embodiment 1: As Figure 1 shown, the electric bicycle charging monitoring method described in this embodiment includes the following processes: Obtain the residential electricity load in real time.

[0019] Perform wavelet packet decomposition on the residential electricity load to obtain the transient load of the high-frequency signal, and locate the load power consumption event based on the transient load of the high-frequency signal.

[0020] Obtain the steady-state load data of several cycles before and after the time point of the load power consumption event, calculate a single load sample based on the steady-state load data, and extract the feature set of the single load sample.

[0021] After clustering the feature set of the single load sample, input it into the pre-trained LightGBM decision tree classifier, and the LightGBM decision tree classifier outputs the electric bicycle charging monitoring result.

[0022] Embodiment 2: As Figure 2 shown, the non-intrusive real-time monitoring method of electric bicycle charging based on FCM (Fuzzy-c means) clustering and LightGBM (Light Gradient Boosting Machine) decision tree classifier for monitoring electric bicycle charging into the household and reducing the incidence of residential fires includes the following steps: S1, Set voltage and current sampling devices at the meter end outside the residential household to monitor the residential electricity load in real time.

[0023] S2, Perform wavelet packet decomposition on the collected load current and voltage to extract waveform features of different frequency bands, and preliminarily judge whether a load power consumption event occurs according to the transient load mutation of the high-frequency signal. If it appears, locate the load power consumption event point. If not, continue to monitor.

[0024] S3. For several cycles before and after the time point of the load power consumption event, calculate a single load sample and extract the feature set of the single load sample.

[0025] S4. Input the feature set of the single load sample into the LightGBM decision tree classifier that has completed training and convergence, and determine whether the monitored load charging behavior is an electric bicycle charging event indoors.

[0026] S5. If the output result determines it is an electric bicycle charging event indoors, report this illegal charging event to the operator via 4G (the fourth-generation mobile communication standard) wireless communication, and conduct subsequent investigation of illegal charging; if the judgment result is not an electric bicycle charging event indoors, record and store this load power consumption information but do not report it.

[0027] S6. Store the labeled sample data for the update and optimization of the LightGBM decision tree classifier.

[0028] The specific content of the above-mentioned S1 is as follows: By installing a sampling device at the electricity meter end outside the residential household, the voltage signal and current signal of the residential electricity load can be obtained in real time. The sampling device does not need to be directly connected to each electrical device, only needs to monitor the total power signal, and is convenient to install and has little interference. The sampling device includes a current transformer, a voltage sensor, and a high-sampling-rate analog-to-digital converter.

[0029] The specific content of the above-mentioned S2 is as follows: Attach an edge computing device to the sampling device end, which can be an MCU single-chip microcomputer, to process and analyze the collected voltage and current data in real time. The processing and analysis include wavelet packet decomposition feature extraction and load power consumption event positioning.

[0030] Among them, wavelet packet decomposition is suitable for analyzing non-stationary signals with complex frequency characteristics, such as the current signals of electric bicycle loads and household loads, which helps to obtain more detailed information about the signals.

[0031] Assume that the original load current signal is I0, and the decomposed signal of the j-th node in the k-th layer is I k,j , and the low-frequency and high-frequency signals of the (k + 1)-th layer at this point are expressed by the following formula: (1) In the formula, h ( m- 2 n ) is the low-frequency signal filtering coefficient; g ( m- 2 n ) is the high-frequency signal filtering coefficient; N is the filter length, and for the wavelet function db4, N = 4; I k+1,2j is the low-frequency signal component;I k+1，2j+1 is a high-frequency signal component, n is the nth sampling point, and m is the mth sampling point.

[0032] The two-layer wavelet packet decomposition results of the transient load and steady-state load of the electric bicycle are shown in the appendix Figure 3 . When the event of an electric bicycle charging at home occurs, it will cause two sudden changes in the current, and the high-frequency signal of the transient load will also have two sudden changes. By analyzing the sudden changes in the high-frequency signal, the electric bicycle load charging event can be reflected, such as Figure 3 in the frequency band of 375 - 500 Hz corresponding to node 6.

[0033] The wavelet packet decomposition result of the steady-state load is taken as part of the characteristics of a single load sample. The energies S 2,0 、S 2,1 、S 2,2 、S 2,3 of different frequency bands obtained by the two-layer wavelet packet decomposition of the current are selected to better capture the distribution of current energies in different frequency bands of the steady-state load of the electric bicycle for electric bicycle load identification. The calculation of the energies of different frequency bands obtained by the two-layer wavelet packet decomposition is as follows: (2) where: I 2,j represents the component of the jth node in the second layer of the wavelet packet decomposition; S 2,j represents the energy of the jth node in the second layer of the wavelet packet decomposition.

[0034] The specific value of S3 is as follows: After determining the occurrence point of the load power consumption event, the steady-state load data of the periods before and after the event point are taken, and the single load sample of the steady-state load data pair is calculated according to the superposition principle. The characteristics of the single load sample are extracted through Fourier transform and wavelet packet decomposition, including instantaneous power, active power, THD (Total Harmonic Distortion), power factor, root mean square current, current peak value, direct current component of current, fundamental component, 3rd harmonic, 5th harmonic, and 7th harmonic, as well as the energies S 2,0 、S 2,1 、S 2,2 、S 2,3 . The above characteristics form the characteristic set of a single load sample for electric bicycle load judgment.

[0035] The specific value of S4 is as follows: The LightGBM decision tree classifier is pre-trained in advance through the training set.

[0036] To ensure the rapid convergence of the training of the LightGBM decision tree classifier, before the training of the LightGBM decision tree classifier, feature screening is performed on the feature set of the single-load samples extracted in S4), and the dominant features are extracted. The feature screening is realized by FCM clustering. The clustering results of all feature combinations in the feature set of the single-load samples are directly evaluated by the clustering accuracy rate, and the feature combination with the highest accuracy rate is selected as the dominant feature for the subsequent training of the LightGBM decision tree classifier. The accuracy rate A cc is: (3) In the formula: T P 、T N 、F P 、F N are the true positive, true negative, false positive, and false negative respectively.

[0037] The FCM clustering result of the finally selected dominant feature combination is shown in Appendix Figure 4 , and the clustering accuracy rate of this feature combination is 0.9898. The corresponding dominant features are the power factor, THD, DC component of the current, 5th harmonic, and S 2,1 of the second-layer decomposition of the wavelet packet.

[0038] After screening out the dominant features through FCM clustering, the dominant features are used as the training set for the training of the LightGBM decision tree classifier. The maximum number of leaf nodes of each decision tree is 31, the convergence accuracy is 1e-6, the number of iterations is 100 times, and two new decision trees are generated in each iteration. The final classifier result is determined by 200 decision trees.

[0039] The LightGBM decision tree classifier includes the calculation and iteration of the loss function and the objective function.

[0040] Loss function: used to measure the gap between the true value and the predicted value. Suppose the training set X is an n×m matrix, where n rows are the number of samples and m is the features included in the samples. Y is the label value of each sample, which is a one-dimensional array with n rows. For the classification problem, the loss function is expressed by the following formula: (4) In the formula: L is the loss function, y i and are the true label corresponding to the i-th sample and the predicted value of the training set sample by the LightGBM decision tree classifier respectively.

[0041] Objective function: The optimization objective of the LightGBM decision tree classifier is to minimize the loss function, as shown in the following formula: (5) (6) In the formula: F represents the set of prediction function parameters of the LightGBM decision tree classifier, which is a set composed of a series of decision trees t; is the regularization term of the objective function, which is used to control the complexity of the LightGBM decision tree classifier and prevent overfitting; T is the number of decision trees; w k is the k weight of the leaf node of the γ th number; λ and

[0042] are hyperparameters that control the regularization strength. (7) In the formula: G L and H L are the g sum and h sum of the left node respectively; G R and H R are the g sum and h sum of the right node respectively; g represents the gradient of the loss function with respect to the predicted value of the LightGBM decision tree classifier; h is the second derivative of the loss function with respect to the predicted value of the LightGBM decision tree classifier. Gain can be regarded as an index indicating whether a leaf node should be split. If it is greater than 0, continue to split; otherwise, stop splitting. This continues until the loss cost is less than the threshold or the maximum number of iterations is reached.

[0043] After the LightGBM decision tree classifier is trained, the dominant features of any household load or electric bicycle load are extracted and used as the input of the LightGBM decision tree classifier. The output of the LightGBM decision tree classifier is whether there is an event of electric bicycle charging indoors. If there is an event of electric bicycle charging indoors, the LightGBM decision tree classifier outputs 1; if there is no event of electric bicycle charging indoors, the LightGBM decision tree classifier outputs 0.

[0044] Specifically, S6 is as follows: According to the output results of the LightGBM decision tree classifier, the dominant features are stored as labeled data for the update and optimization of the LightGBM decision tree classifier.

[0045] Household loads and electric bicycle loads were randomly selected for accuracy verification, and the results are shown in Table 1. In the table, P is the precision; R is the recall rate; F1 is the F1 score, which is used to evaluate the adaptability of the LightGBM decision tree classifier to unbalanced samples. The identification accuracy of all user loads is not less than 0.997, and the F1 score is not less than 0.995, indicating that the LightGBM decision tree classifier has good adaptability for dealing with problems such as electric bicycle load identification.

[0046] Table 1 Accuracy Verification

[0047] Example 3: In this embodiment, an electric bicycle charging monitoring system is provided. This electric bicycle charging monitoring system can be used to implement the above-mentioned electric bicycle charging monitoring method. Specifically, the electric bicycle charging monitoring system includes a load acquisition module, a load power consumption event positioning module, a feature extraction module, and a result output module.

[0048] Among them, the load acquisition module is used to acquire the residential electricity load in real time.

[0049] The load power consumption event positioning module is used to perform wavelet packet decomposition on the residential electricity load to obtain the transient load of the high-frequency signal, and locate the load power consumption event based on the transient load of the high-frequency signal.

[0050] The feature extraction module is used to obtain the steady-state load data of several cycles before and after the time point of the load power consumption event, calculate a single load sample based on the steady-state load data, and extract the feature set of the single load sample.

[0051] The result output module is used to cluster the feature set of the single load sample and then input it into the pre-trained LightGBM decision tree classifier, and the LightGBM decision tree classifier outputs the electric bicycle charging monitoring result.

[0052] Example 4: In this embodiment, a terminal device is provided. The terminal device includes a processor and a memory. The memory is used to store a computer program, and the computer program includes program instructions. The processor is used to execute the program instructions stored in the computer storage medium. The processor may be a Central Processing Unit (CPU), or may also be other general-purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field-Programmable Gate Arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing core and control core of the terminal, and is suitable for implementing one or more instructions. Specifically, it is suitable for loading and executing one or more instructions to implement the corresponding method flow or corresponding function. The processor described in the embodiment of the present invention can be used for the operation of the electric bicycle charging monitoring method, including: obtaining the residential electricity load in real time; performing wavelet packet decomposition on the residential electricity load to obtain the transient load of the high-frequency signal, and locating the load electricity event based on the transient load of the high-frequency signal; obtaining the steady-state load data of several cycles before and after the time point of the load electricity event, calculating a single load sample based on the steady-state load data, and extracting the feature set of the single load sample; after clustering the feature set of the single load sample, inputting it into a pre-trained LightGBM decision tree classifier, and the LightGBM decision tree classifier outputs the electric bicycle charging monitoring result.

[0053] Embodiment 5: In this embodiment, a computer-readable storage medium (Memory) is provided. The computer-readable storage medium is the memory device in the terminal device and is used to store programs and data. It can be understood that the computer-readable storage medium here can include both the built-in storage medium in the terminal device and, of course, the extended storage medium supported by the terminal device. The computer-readable storage medium provides a storage space, and this storage space stores the operating system of the terminal. And, one or more instructions suitable for being loaded and executed by the processor are also stored in this storage space. These instructions can be one or more computer programs (including program codes). It should be noted that the computer-readable storage medium here can be a high-speed Random Access Memory (RAM), or a non-volatile memory, such as at least one disk memory.

[0054] One or more instructions stored in a computer-readable storage medium can be loaded and executed by a processor to implement the corresponding steps of the electric bicycle charging monitoring method in the above embodiments; one or more instructions in the computer-readable storage medium are loaded and executed by the processor to perform the following steps: obtain the residential electricity load in real time; perform wavelet packet decomposition on the residential electricity load to obtain the transient load of the high-frequency signal, and locate the load electricity event based on the transient load of the high-frequency signal; obtain the steady-state load data of several cycles before and after the time point of the load electricity event, calculate a single load sample based on the steady-state load data, and extract the feature set of the single load sample; after clustering the feature set of the single load sample, input it into a pre-trained LightGBM decision tree classifier, and the LightGBM decision tree classifier outputs the electric bicycle charging monitoring result.

[0055] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, optical storage, etc.) containing computer-usable program code.

[0056] The present application is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram can be implemented by computer program instructions, and the combination of the flows and / or blocks in the flowchart and / or block diagram can also be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate a device for implementing the specified functions in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0057] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured article including an instruction device, and the instruction device implements the specified functions in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0058] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus, causing a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, so that the instructions executed on the computer or other programmable apparatus provide steps for implementing the functions specified in one process or a plurality of processes and / or blocks Figure 1 one process or a plurality of processes and / or blocks Figure 1 or steps for implementing the functions specified in a plurality of blocks or a plurality of blocks.

[0059] In the above embodiments of the present application, the descriptions of the various embodiments have their respective focuses. For parts not detailed in a certain embodiment, reference may be made to the relevant descriptions of other embodiments.

[0060] The above are only the preferred embodiments of the present application. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present application, several improvements and modifications can be made, and these improvements and modifications should also be regarded as the protection scope of the present application.

[0061] It should be understood that the above description is for the purpose of illustration and not limitation. Many embodiments and many applications other than the examples provided will be apparent to those skilled in the art upon reading the above description.

Claims

1. An electric bicycle charging monitoring method, characterized in that, It includes the following processes: Obtain the residential electricity load in real time; Perform wavelet packet decomposition on the residential electricity load to obtain the transient load of the high-frequency signal, and locate the load power consumption event based on the transient load of the high-frequency signal; Obtain the steady-state load data of several cycles before and after the time point of the load power consumption event, calculate a single load sample based on the steady-state load data, and extract the feature set of the single load sample; After clustering the feature sets of the single load samples, input them into a pre-trained LightGBM decision tree classifier, and the LightGBM decision tree classifier outputs the electric bicycle charging monitoring result.

2. The electric bicycle charging monitoring method according to claim 1, wherein The specific process of performing wavelet packet decomposition on the residential electricity load to obtain the transient load of the high-frequency signal and locating the load power consumption event based on the transient load of the high-frequency signal is as follows: Perform multi-layer wavelet packet decomposition on the load current in the residential electricity load to obtain the transient load of the high-frequency signal. If the transient load of the high-frequency signal changes suddenly, a load power consumption event occurs.

3. The electric bicycle charging monitoring method according to claim 1, wherein The specific process of calculating a single load sample based on the steady-state load data and extracting the feature set of the single load sample is as follows: Calculate the single load sample of the steady-state load data of several cycles according to the superposition principle, and extract the feature set of the single load sample through Fourier transform and wavelet packet decomposition.

4. The electric bicycle charging monitoring method according to claim 1, characterized in that, The feature set of the single load sample includes: instantaneous power, active power, THD, power factor, effective current value, current peak value, DC component of current, fundamental component, 3rd harmonic, 5th harmonic, 7th harmonic, and the energy of different frequency bands decomposed by wavelet packet at the second layer.

5. The electric bicycle charging monitoring method according to claim 1, characterized in that, The training process of the LightGBM decision tree classifier is as follows: Use the feature set of the single load sample as the training set to train the LightGBM decision tree classifier, and set the maximum number of leaf nodes, convergence accuracy, and number of iterations; Optimize the LightGBM decision tree classifier by minimizing the loss function to obtain the trained LightGBM decision tree classifier.

6. The electric bicycle charging monitoring method according to claim 5, characterized in that, The specific process of clustering the feature sets of the single load samples is as follows: Screen the feature sets of the single load samples through FCM clustering, extract the dominant features, and use the dominant features as the training set.

7. The electric bicycle charging monitoring method according to claim 6, wherein The specific process of screening the feature sets of the single load samples through FCM clustering and extracting the dominant features is as follows: Calculate the clustering accuracy of different feature combinations in the feature set of the single load sample; Select the feature combination with the highest clustering accuracy as the dominant feature.

8. An electric bicycle charging monitoring system, characterized in that, It includes: A load acquisition module for obtaining the residential electricity load in real time; A load power consumption event location module for performing wavelet packet decomposition on the residential electricity load to obtain the transient load of the high-frequency signal and locating the load power consumption event based on the transient load of the high-frequency signal; A feature extraction module for obtaining the steady-state load data of several cycles before and after the time point of the load power consumption event, calculating a single load sample based on the steady-state load data, and extracting the feature set of the single load sample; A result output module for clustering the feature sets of the single load samples and then inputting them into a pre-trained LightGBM decision tree classifier, and the LightGBM decision tree classifier outputs the electric bicycle charging monitoring result.

9. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, the steps of the electric bicycle charging monitoring method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, the steps of the electric bicycle charging monitoring method according to any one of claims 1 to 7 are implemented.

Images