Load waveform feature analysis and complexity classification and evaluation method and system for electrical equipment

Abstract

This invention discloses a method and system for analyzing the waveform characteristics and classifying the complexity of electrical equipment loads. The method includes data acquisition. M The instantaneous current of each load sample is used for zero-crossing detection and alignment selection for subsequent continuous loads. T The instantaneous current at each sampling point; for all load samples, respectively according to T The instantaneous current at each sampling point is used to calculate various sample characteristics; after normalizing the sample characteristics of all load samples, clustering is performed; for each load sample, the normalized sample characteristics of that load sample are calculated in the clustering results. Z The total weighted distance across features is used to determine the waveform score and equipment complexity of the load sample through a pre-defined waveform scoring model. This invention aims to quantify the complexity of load waveforms of electrical equipment, eliminate model dependence, achieve cross-method and cross-equipment performance evaluation, and ensure the objective comparability of evaluation results based on different non-intrusive load monitoring and identification methods.

Description

Technical Field

[0001] This invention belongs to the field of power system analysis and smart grid technology, specifically relating to a method and system for analyzing the waveform characteristics and classifying the complexity of electrical equipment loads. Background Technology

[0002] Non-intrusive load monitoring (NILM) is a method for identifying the status and analyzing the energy consumption of power equipment based on electrical signal characteristics such as voltage, current, and power. NILM technology enables the classification, operational status detection, and energy consumption statistics of electrical equipment such as household appliances and industrial loads without the need for additional sensors, using only measuring devices such as smart meters. Therefore, a comprehensive and scientific evaluation of NILM algorithms is crucial. Currently, the main challenges in evaluating NILM algorithms are the difficulty in acquiring load data across multiple scenarios and the reliance on specific models for evaluation results. First, the power range (from a few watts to hundreds of kilowatts) and operating modes (intermittent and continuous) of load data required for NILM tasks vary significantly. Data from a single device cannot cover all scenario characteristics, and there are many combinations of different devices, making the acquisition of data from specific device combinations difficult. Second, existing methods mainly use machine learning to identify loads, and the performance of the algorithm largely depends on the identification results. Furthermore, different algorithms use different evaluation metrics, and comparing only the same evaluation metrics cannot guarantee fairness. Summary of the Invention

[0003] The technical problem to be solved by this invention is to provide a method and system for analyzing the characteristics and classifying the complexity of load waveforms of electrical equipment, thereby eliminating model dependence and enabling performance evaluation across methods and devices, and ensuring that the evaluation results based on different non-intrusive load monitoring and identification methods are objectively comparable.

[0004] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows:

[0005] A method for analyzing the waveform characteristics and classifying the complexity of electrical equipment loads includes the following steps:

[0006] S1, data collection M The instantaneous current of each load sample;

[0007] S2, respectively for M Zero-crossing detection is performed on the instantaneous current of each load sample, and the first detected zero-crossing position is used to align and select subsequent continuous load samples. T The instantaneous current at each sampling point;

[0008] S3, for all load samples, respectively according to TThe instantaneous current at each sampling point is used to calculate various sample characteristics;

[0009] S4, clustering is performed after normalizing the sample features of all load samples;

[0010] S5, for each load sample, calculate the normalized sample features of that load sample in the clustering results. Z The total weighted distance across all features;

[0011] S6, for each load sample, the waveform score and electrical equipment complexity are determined by a preset waveform scoring model based on the total weighted distance.

[0012] Optionally, in step S2, respectively... M Zero-crossing detection of the instantaneous current of each load sample includes:

[0013] S2.1, calculates the sign change difference for the instantaneous current of each load sample:

[0014] ;

[0015] ;

[0016] in, for Sign change difference, For load samples m No. n Instantaneous current at each sampling point For symbolic functions, For load samples m No. n -1 instantaneous current at sampling points;

[0017] S2.2, determine the set of zero-crossing indices for each load sample according to the following formula:

[0018] ;

[0019] in, For load samples m The set of zero-crossing indices Indicates load sample m No. u A zero-crossing position index, ;

[0020] S2.3, determine the starting point of the instantaneous current for each load sample based on the zero-crossing index set:

[0021] ;

[0022] in, For load samples m The starting point of the instantaneous current, To obtain the minimum value.

[0023] Optionally, in step S3, according to T When calculating multiple sample characteristics for the instantaneous current of a sampling point, the multiple sample characteristics of each load sample include some or all of the fundamental amplitude, harmonic amplitude of a specified order, current peak value, current peak-to-peak value, crest factor, and harmonic distortion rate of the instantaneous current of that load sample.

[0024] Optionally, the specified harmonic amplitude includes some or all of the amplitudes of the 3rd, 5th, 7th, and 9th harmonics.

[0025] Optionally, the clustering of all load samples after normalizing their sample features in step S4 includes:

[0026] S4.1, normalize the sample characteristics of all load samples;

[0027] S4.2, Obtain the determined number of cluster centers P z The K-Means++ algorithm is used to determine the initial cluster center vectors. :

[0028] ;

[0029] in, ~ They are respectively number 1 to P z The initial cluster centers of each cluster;

[0030] S4.3, the cluster centers of each cluster are iteratively updated according to the following formula:

[0031] ;

[0032] ;

[0033] ;

[0034] in, For the (t+1)th iteration, the... p The cluster centers of each cluster; For sample set The number of samples; Let be the sample set, representing the set of samples in the first case. t In the next iteration, it was classified as the... A collection of device indexes for a class; Normalized load sample m Thez Individual sample features; The optimal category index is an integer label representing the payload sample. m The z Each sample feature sample The category being classified;

[0035] S4.4 Determine whether the termination condition of the following expression is true:

[0036]

[0037] in, For the t-th iteration, the th... p The cluster centers of each cluster. Set a preset threshold; if the termination condition is not met, proceed to step S4.3 to continue iterative updates; otherwise, proceed to step S5.

[0038] Optionally, in step S5, for each load sample, the normalized sample features of that load sample are calculated in the clustering results. Z The functional expression for the total weighted distance on each feature is:

[0039] ;

[0040] in, For load samples m Total weighted distance M The total number of load samples, w z For the first z Weighting coefficients for each sample feature, Normalized load sample m The z Individual sample features, For load samples m The z Cluster centers corresponding to the features of each sample.

[0041] Optionally, step S6 includes:

[0042] S6.1, For each load sample, the waveform score is determined using a preset waveform scoring model based on the total weighted distance.

[0043] ;

[0044] ;

[0045] in, For load samples m Waveform score, For load samplesm Total weighted distance The median of the waveform score. Interquartile range, This is a very small constant used to prevent the denominator from being zero; median For median operations, ~ These represent the total weighted distances from load sample 1 to load sample M, respectively.

[0046] S6.2, For the waveform scoring of load samples 1 to load samples M, the classification is based on quantiles:

[0047] ;

[0048] in, It is the first quartile, and it is the set. The scores that rank in the top 25% The median is , and the set is . The scores in the top 50% of the class, The third quartile is a set The top 75% of scores, including the set This is a set of waveform scores for load samples 1 to M; Class A, Class B, Class C, and Class D waveforms represent the complexity of electrical equipment from high to low, with higher complexity indicating more types and quantities of electrical equipment.

[0049] The present invention also provides a system for analyzing and classifying the load waveform characteristics and complexity of electrical equipment, comprising a microprocessor and a memory interconnected thereto, wherein the microprocessor is programmed or configured to execute the method for analyzing and classifying the load waveform characteristics and complexity of electrical equipment.

[0050] The present invention also provides a computer-readable storage medium storing a computer program or instructions that are programmed or configured to execute the electrical equipment load waveform characteristic analysis and complexity classification assessment method by a processor.

[0051] The present invention also provides a computer program product, including a computer program or instructions, which are programmed or configured to execute the electrical equipment load waveform characteristic analysis and complexity classification evaluation method by a processor.

[0052] Compared with existing technologies, this invention mainly achieves the following beneficial effects: 1. Load data acquisition typically relies on physical measurement devices (such as smart meters and data acquisition terminals). However, these devices are affected by factors such as the measurement environment, device deployment location, and sampling frequency, making it difficult to obtain diverse load operation waveforms under different scenarios. This invention synthesizes load operation waveforms under target scenarios using the instantaneous current of specific devices, simulating load operation waveforms under different device combinations and overcoming the limitations of traditional data acquisition. 2. Existing load identification and evaluation methods rely on the identification results of machine learning models for performance evaluation. However, the inconsistent input data for model training affects the reliability of the evaluation results, making it difficult to establish a unified evaluation method across different datasets and algorithms. This invention collects the instantaneous current of the load, calculates the harmonics, peak-to-peak values, distortion rate, and other features of the current, and uses K-means clustering to calculate the distance between different features and feature centers, thereby evaluating the complexity of the load waveform. This does not rely on a specific load identification model, making the evaluation results of different identification methods comparable. This invention can quantify the complexity of load waveforms of electrical equipment, eliminate model dependence, realize cross-method and cross-equipment performance evaluation, and ensure that the evaluation results based on different non-intrusive load monitoring and identification methods are objectively comparable. Attached Figure Description

[0053] Figure 1 This is a schematic diagram of the basic process of the method in an embodiment of the present invention.

[0054] Figure 2 This is the harmonic spectrum of the actual load sample when the kettle and air conditioner are running simultaneously in an embodiment of the present invention.

[0055] Figure 3 This is the harmonic spectrum of the synthesized load sample when the kettle and the air conditioner are running simultaneously in an embodiment of the present invention.

[0056] Figure 4 The waveforms of the composite load sample when the kettle and the air conditioner are running simultaneously in the embodiment of the present invention are shown below. (a) is the instantaneous current when the kettle is running stably, (b) is the instantaneous current when the air conditioner is running stably in cooling mode, (c) is the instantaneous current when the kettle and the air conditioner are running simultaneously in cooling mode, and (d) is the composite instantaneous current when the kettle and the air conditioner are running simultaneously in cooling mode. Detailed Implementation

[0057] To enable those skilled in the art to better understand the technical solutions of the present invention, the technical solutions of the present invention will be further described in detail below with reference to the accompanying drawings in the embodiments of the present invention.

[0058] like Figure 1 As shown, the method for analyzing the waveform characteristics and classifying the complexity of electrical equipment loads in this embodiment includes the following steps:

[0059] S1, data collection M The instantaneous current of each load sample;

[0060] S2, respectively for M Zero-crossing detection is performed on the instantaneous current of each load sample, and the first detected zero-crossing position is used to align and select subsequent continuous load samples. T The instantaneous current at each sampling point;

[0061] S3, for all load samples, respectively according to T The instantaneous current at each sampling point is used to calculate various sample characteristics;

[0062] S4, clustering is performed after normalizing the sample features of all load samples;

[0063] S5, for each load sample, calculate the normalized sample features of that load sample in the clustering results. Z The total weighted distance across all features;

[0064] S6, for each load sample, the waveform score and electrical equipment complexity are determined by a preset waveform scoring model based on the total weighted distance.

[0065] In step S1 of this embodiment, data is collected. M When the instantaneous current of each load sample is measured, including at a sampling rate... f s collection M The instantaneous current data matrix I, obtained from the stable operation of each device, constitutes the data matrix:

[0066] ;

[0067] in, ~ Let I represent the instantaneous current at the first sampling point of load sample 1 to the instantaneous current at the Nth sampling point of load sample M, respectively. M × N The matrix, N The total number of sampling points for each device, and In traditional load identification and evaluation, the sample to be tested cannot cover all operating scenarios of the equipment, resulting in insufficient reliability of the evaluation results. Given the high cost of comprehensive sample collection, this embodiment includes load samples of single devices and combinations of multiple devices. By synthesizing the current of multiple devices operating in combination, it simulates the scenario of multiple devices operating simultaneously, supplementing different operating conditions and improving the scientific rigor and reliability of the evaluation method. Assumptions are selected... l Current synthesis when multiple devices are operating in combination. The instantaneous current of the combined multi-device system during operation is defined as follows: :

[0068] ;

[0069] By collecting load samples including single devices and combinations of multiple devices, the versatility of the electrical equipment load waveform characteristic analysis and complexity classification assessment method in this embodiment can be improved. Specifically, the number of sampling points for each load sample... N When the value is 1,920,000, then:

[0070] ;

[0071] Specifically, in this embodiment, the instantaneous current I of five devices operating stably for five minutes is collected at a sampling rate of 6400Hz to form a data matrix:

[0072] ;

[0073] In the formula, I is a matrix of size 5×1920000, and the number of sampling points for each load sample is... N It is 1,920,000.

[0074] In step S2 of this embodiment, respectively... M The instantaneous current of each load sample is zero-crossing detected to obtain the aligned instantaneous current of different load samples, including:

[0075] S2.1, calculates the sign change difference for the instantaneous current of each load sample:

[0076] ;

[0077] ;

[0078] in, for Sign change difference, For load samples m No. n Instantaneous current at each sampling point For symbolic functions, For load samples m No. n -1 instantaneous current at sampling points;

[0079] S2.2, determine the set of zero-crossing indices for each load sample according to the following formula:

[0080] ;

[0081] in, For load samples m The set of zero-crossing indices Indicates load sample m No. u A zero-crossing position index, The specific number of sampling points for each load sample. N When the value is 1,920,000, then:

[0082] ;

[0083] S2.3, determine the starting point of the instantaneous current for each load sample based on the zero-crossing index set:

[0084] ;

[0085] in, For load samples m The starting point of the instantaneous current, To minimize the value, an aligned instantaneous current data matrix can be constructed by aligning all load samples. :

[0086] ;

[0087] in, ~ Select subsequent consecutive values ​​for each load sample alignment. T The instantaneous current vector at each sampling point ~ This represents the instantaneous current of N sampling points after aligning load samples m. Taking five load samples as an example, the instantaneous current after aligning the five devices can be expressed as:

[0088] ;

[0089] From the aligned instantaneous current data matrix Select subsequent consecutive T The instantaneous current at each sampling point can be expressed as:

[0090] ;

[0091] in, Aligned selection for all load samples T Instantaneous current data matrix at each sampling point ~ Aligned for each load sample T The instantaneous current data at each sampling point are as follows:

[0092] ;

[0093] For load samples with multiple devices combined, then... , , T =60* f s , T < N , m ∈{1,2,…, M In this embodiment, the instantaneous current with a magnitude of T=384000 is extracted from the aligned instantaneous current:

[0094] .

[0095] In step S3 of this embodiment, respectively, according to... T When calculating multiple sample characteristics for the instantaneous current at each sampling point, the multiple sample characteristics for each load sample include some or all of the fundamental amplitude, specified harmonic amplitude, peak current, peak-to-peak current, crest factor, and harmonic distortion rate of the instantaneous current of that load sample. The specified harmonic amplitude includes some or all of the amplitudes of the 3rd, 5th, 7th, and 9th harmonics. Based on the instantaneous current of the equipment... The load samples were calculated using the Fast Fourier Transform (FFT). m No. k The amplitude of the second harmonic current component, that is, the fundamental amplitude and the amplitude of the specified harmonic of the instantaneous current of the load sample:

[0096] ;

[0097] in, Let k be the amplitude of the harmonic. , , , , These represent the fundamental frequency amplitude, and the amplitudes of the 3rd, 5th, 7th, and 9th harmonics, respectively. The calculation functions for peak current, peak-to-peak current, crest factor, and harmonic distortion rate are as follows:

[0098] ;

[0099] ;

[0100] ;

[0101] Specifically, the number of sampling points for each load sample N When the value is 1,920,000, then:

[0102]

[0103] .

[0104] in, , , and These represent the peak current, peak-to-peak current, crest factor, and harmonic distortion rate, respectively. Based on this, the sample characteristics of all load samples can be represented as a feature matrix. :

[0105] ;

[0106] in, , , and These represent load samples respectively. m The peak current, peak-to-peak current, crest factor, and harmonic distortion rate.

[0107] In traditional load identification and evaluation, the performance of the algorithm mainly depends on the identification results. However, the method in this embodiment starts from the load waveform itself and uses the K-Means clustering algorithm to find features. z Cluster center, computing device m The total weighted distance from different features to the cluster centers quantifies the complexity of the waveform without relying on any specific load identification model. Specifically, in this embodiment, step S4, clustering after normalizing the sample features of all load samples includes:

[0108] S4.1, Normalize the sample features of all load samples; to eliminate the influence of dimensions, this embodiment normalizes the feature matrix. The normalized function expression is:

[0109] ;

[0110] ;

[0111] ;

[0112] in, Normalized load sample m The z Individual sample features; For load samples m The z Individual sample features, and The first of all load samples z The mean and standard deviation of each sample feature. m =1,2,…, M , z =1,2,…, Z ;

[0113] S4.2, Obtain the determined number of cluster centers P z Determining the number of cluster centers using silhouette coefficient P z And it needs to meet the following requirements P z ≥2, use the K-Means++ algorithm to determine the initial cluster center vector. :

[0114] ;

[0115] in, ~ They are respectively number 1 to P z The initial cluster centers of each cluster;

[0116] S4.3, the cluster centers of each cluster are iteratively updated according to the following formula:

[0117] ;

[0118] ;

[0119] ;

[0120] in, For the (t+1)th iteration, the... p The cluster centers of each cluster; For sample set The number of samples; Let be the sample set, representing the set of samples in the first place. t In the next iteration, it was classified as the... A collection of device indexes for a class; Normalized load sample m The z Individual sample features; The optimal category index is an integer label representing the payload sample. m The z Each sample feature sample The category being classified;

[0121] S4.4 Determine whether the termination condition of the following expression is true:

[0122]

[0123] in, For the t-th iteration, the th... p The cluster centers of each cluster. The preset threshold is used; if the termination condition is not met, proceed to step S4.3 to continue iterative updates; otherwise, proceed to step S5. In this embodiment, the preset threshold is 0.0001, that is, the termination condition is: .

[0124] In step S5 of this embodiment, for each load sample, the normalized sample features of that load sample are calculated in the clustering results. Z The functional expression for the total weighted distance on each feature is:

[0125] ;

[0126] in, For load samples m Total weighted distance M The total number of load samples, For the first z Weighting coefficients for each sample feature, Normalized load sample m The z Individual sample features, For load samples m The z Cluster centers corresponding to each sample feature. Weighting coefficients. It can be configured according to the actual situation, such as taking the weight when equal weighting. =1 / Z .

[0127] In traditional load identification and evaluation, different algorithms use different evaluation metrics. Considering that comparing some of the same evaluation metrics across different algorithms cannot guarantee fairness, this embodiment introduces a waveform scoring model to ensure the objective comparability of evaluation results based on different identification methods. In non-intrusive load identification, loads with similar electrical characteristics are easily confused, i.e., the total weighted distance of characteristics of different devices... The closer the numerical values ​​are, the harder it is to distinguish them. Therefore, in M In a load sample, the equipment waveform that is difficult to distinguish is called a complex waveform, and vice versa. Specifically, step S6 in this embodiment includes:

[0128] S6.1, For each load sample, the waveform score is determined using a preset waveform scoring model based on the total weighted distance.

[0129] ;

[0130] ;

[0131] in, For load samples mWaveform score, For load samples m Total weighted distance The median of the waveform score. Interquartile range, To prevent extremely small constants with denominators of 0, a value of 0.001 is used in this embodiment, so:

[0132] ;

[0133] median For median operations, ~ These represent the total weighted distances from load sample 1 to load sample M; where the interquartile range is... The calculation includes: the total weighted distance for all load samples. Sort in ascending order, and then calculate the quantiles using the following formula:

[0134] ;

[0135] ;

[0136] in, percentile { D , b} represents the quantile operation, applied to sets. D After sorting in ascending order, at the optimal position Take the value at the location, and then according to The interquartile range can then be obtained. To avoid the denominator being 0, when IQR When = 0, add a minimal constant to the denominator. ;

[0137] S6.2, For the waveform scoring of load samples 1 to load samples M, the classification is based on quantiles:

[0138] ;

[0139] in, It is the first quartile, and it is the set. The scores that rank in the top 25% The median is , and the set is . The scores in the top 50% of the class, The third quartile is a set The top 75% of scores, including the set This is a set of waveform scores for load samples 1 to M; waveforms of type A, B, C, and D represent the four levels of electrical equipment complexity from highest to lowest. Higher equipment complexity includes more types and quantities of electrical equipment. It can be represented as:

[0140] ;

[0141] First quartile , median and the third and fourth quartiles The expression for the computation function is:

[0142] ;

[0143] ;

[0144] ;

[0145] The resulting Class A, Class B, Class C, and Class D waveforms represent the four levels of electrical equipment complexity, from highest to lowest. For example, in this embodiment, the Class A waveform is... m Waveforms that are difficult to distinguish in various devices often have low identification accuracy; Class B waveforms are... m The waveforms that are more difficult to distinguish among the various devices, but whose identification accuracy is slightly higher than that of Class A waveforms; Class C waveforms are m The waveforms that are easier to distinguish among the various devices have slightly lower identification accuracy than Class D waveforms; Class D waveforms are... m The waveforms are easily distinguishable among the devices and have high identification accuracy.

[0146] To verify the effectiveness of the electrical equipment load waveform characteristic analysis and complexity classification assessment method in this embodiment, the harmonic spectra of the actual measured current signal and the synthesized current signal when the kettle and air conditioner are running simultaneously are compared. The harmonic spectra of the actual load sample when the kettle and air conditioner are running simultaneously in this embodiment are as follows: Figure 2 As shown in Table 1, the harmonic spectrum of the synthesized load sample when the kettle and air conditioner are running simultaneously in this embodiment is as follows: Figure 3 As shown in Table 2, Figure 2 and Figure 3 In the diagram, H2 to H15 represent the 2nd to 15th harmonics. Figure 4 The waveforms are the composite load samples of the kettle and air conditioner running simultaneously in this embodiment, where (a) is the instantaneous current of the kettle running stably, (b) is the instantaneous current of the air conditioner running stably in cooling mode, (c) is the instantaneous current of the kettle and air conditioner running in cooling mode, and (d) is the composite instantaneous current of the kettle and air conditioner running simultaneously in cooling mode.

[0147] Table 1: Harmonic Spectrum of Instantaneous Current When a Kettle and an Air Conditioner are Running Simultaneously

[0148]

[0149] Table 2: Harmonic spectra of the synthesized instantaneous current.

[0150]

[0151] according to Figure 2 and Figure 3 A comparison of the actual and synthesized current signals, as well as between Tables 1 and 2, reveals that when the kettle and air conditioner are running simultaneously, the fundamental frequencies of the measured and synthesized current signals are identical. The amplitudes of the main low-order harmonic components are similar, and the distribution patterns of the higher-order harmonics are also similar. This indicates that the synthesized waveform retains the frequency structure characteristics of the original signal well. Furthermore, the total harmonic distortion (THD) of the actual waveform is 4.72%, while that of the synthesized waveform is 4.92%, a difference of only 0.20%. This further verifies the accuracy and effectiveness of the synthesis method in the frequency domain, demonstrating its ability to realistically reproduce the current characteristics of multiple electrical appliances operating simultaneously. To verify the effectiveness of the load waveform feature analysis and complexity classification assessment method for electrical equipment in this embodiment, five common electrical appliances were selected for load waveform evaluation. The sampling rate for each appliance was 6400 Hz, and the acquisition time was 5 minutes. The synthesized waveforms were those of a hair dryer and a rice cooker running stably at the same time. The experimental environment configuration was as follows: operating system: Ubuntu 18.04; processor: AMD Ryzen 7 5800H; memory: 40 GB; hard disk capacity: 1 TB; graphics card: NVIDIA GeForce RTX 3060 laptop graphics card with 6 GB of video memory. The development language used is Python, and the development framework is PyTorch. The final device complexity of the load samples corresponding to various devices is as follows: Kettle_High Frequency 3: Device complexity is Class D waveform; Water Heater_High Frequency 1: Device complexity is Class D waveform; Rice Cooker_High Frequency 1: Device complexity is Class C waveform; Hair Dryer_High Frequency 1: Device complexity is Class C waveform; Washing Machine_High Frequency 1: Device complexity is Class B waveform; Composite waveform: Device complexity is Class A waveform. Among these, Kettle_High Frequency 3 refers to a specific operating condition of the kettle. The naming meanings of the other load samples are similar and will not be listed here. The above results verify that the device load waveform feature analysis and complexity classification evaluation method of this embodiment can accurately distinguish between single devices and composite waveforms, quantify the complexity of device load waveforms, eliminate model dependence, achieve cross-method and cross-device performance evaluation, and ensure the objective comparability of evaluation results based on different non-intrusive load monitoring and identification methods.

[0152] Furthermore, this embodiment also provides a system for analyzing and classifying the waveform characteristics and complexity of electrical equipment loads, including a microprocessor and a memory interconnected thereto. The microprocessor is programmed or configured to execute the method for analyzing and classifying the waveform characteristics and complexity of electrical equipment loads. This embodiment also provides a computer-readable storage medium storing a computer program or instructions programmed or configured to execute the method for analyzing and classifying the waveform characteristics and complexity of electrical equipment loads via a processor. This embodiment also provides a computer program product, including a computer program or instructions programmed or configured to execute the method for analyzing and classifying the waveform characteristics and complexity of electrical equipment loads via a processor.

[0153] Those skilled in the art will understand that the technical solutions provided by this invention may take the form of a method, system, or computer program product. Therefore, this invention may take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this invention may take the form of a computer program product embodied on one or more computer-readable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, produce an implementation of the flowchart... Figure 1 One or more processes and / or boxes Figure 1 The computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The functions specified in one or more boxes. These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable apparatus for implementing the process. Figure 1 One or more processes and / or boxes Figure 1The steps of the function specified in one or more boxes.

[0154] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A method for analyzing the waveform characteristics and classifying the complexity of electrical equipment loads, characterized in that, Includes the following steps: S1, data collection M The instantaneous current of a load sample, wherein the load sample includes single device and multi-device combination; S2, respectively for M The instantaneous current of each load sample is zero-crossing detected, and the subsequent continuous loads are selected based on the first detected zero-crossing position. T The instantaneous current at each sampling point; S3, for all load samples, respectively according to T The instantaneous current at each sampling point is used to calculate various sample characteristics; S4, clustering is performed after normalizing the sample features of all load samples; S5, for each load sample, calculate the normalized sample features of that load sample in the clustering results. Z The total weighted distance across all features; S6, for each load sample, the waveform score and equipment complexity are determined using a preset waveform scoring model based on the total weighted distance. This includes: S6.1, for each load sample, the waveform score is determined using a preset waveform scoring model based on the total weighted distance. ； ； in, For load samples m Waveform score, For load samples m Total weighted distance The median of the waveform score. Interquartile range, This is a very small constant used to prevent the denominator from being zero; median For median operations, ~ S6.2 represents the total weighted distance from load sample 1 to load sample M; S6.2 classifies the waveform scores for load samples 1 to load sample M based on quantiles. ； in, It is the first quartile, and it is the set. The scores that rank in the top 25% The median is , and the set is . The scores in the top 50% of the class, The third quartile is a set The top 75% of scores, including the set This is a set of waveform scores for load samples 1 to M; Class A, Class B, Class C, and Class D waveforms represent the complexity of electrical equipment from high to low, with higher complexity indicating more types and quantities of electrical equipment.

2. The method for analyzing the waveform characteristics and classifying the complexity of electrical equipment loads according to claim 1, characterized in that, In step S2, respectively for M Zero-crossing detection of the instantaneous current of each load sample includes: S2.1, calculates the sign change difference for the instantaneous current of each load sample: ； ； in, for Sign change difference, For load samples m No. n Instantaneous current at each sampling point For symbolic functions, For load samples m No. n -1 instantaneous current at sampling points; S2.2, determine the set of zero-crossing indices for each load sample according to the following formula: ； in, For load samples m The set of zero-crossing indices Indicates load sample m No. u A zero-crossing position index, ; S2.3, determine the starting point of the instantaneous current for each load sample based on the zero-crossing index set: ； in, For load samples m The starting point of the instantaneous current, To obtain the minimum value.

3. The method for analyzing the waveform characteristics and classifying the complexity of electrical equipment loads according to claim 1, characterized in that, In step S3, according to T When calculating multiple sample characteristics for the instantaneous current of a sampling point, the multiple sample characteristics of each load sample include some or all of the fundamental amplitude, harmonic amplitude of a specified order, current peak value, current peak-to-peak value, crest factor, and harmonic distortion rate of the instantaneous current of that load sample.

4. The method for analyzing the waveform characteristics and classifying the complexity of electrical equipment loads according to claim 3, characterized in that, The specified harmonic amplitude includes some or all of the amplitudes of the 3rd, 5th, 7th, and 9th harmonics.

5. The method for analyzing the waveform characteristics and classifying the complexity of electrical equipment loads according to claim 1, characterized in that, Step S4 involves clustering all load samples after normalizing their sample features, including: S4.1, normalize the sample characteristics of all load samples; S4.2, Obtain the determined number of cluster centers P z The K-Means++ algorithm is used to determine the initial cluster center vectors. : ； in, ~ They are respectively number 1 to P z The initial cluster centers of each cluster; S4.3, the cluster centers of each cluster are iteratively updated according to the following formula: ； ； ； in, For the (t+1)th iteration, the... p The cluster centers of each cluster; For sample set The number of samples; Let be the sample set, representing the set of samples in the first place. t In the next iteration, it was classified as the... A collection of device indexes for a class; Normalized load sample m The z Individual sample features; The optimal category index is an integer label representing the payload sample. m The z Each sample feature sample The category being classified; S4.4 Determine whether the termination condition of the following expression is true: in, For the t-th iteration, the th... p The cluster centers of each cluster. Set a preset threshold; if the termination condition is not met, proceed to step S4.3 to continue iterative updates; otherwise, proceed to step S5.

6. The method for analyzing the waveform characteristics and classifying the complexity of electrical equipment loads according to claim 1, characterized in that, In step S5, for each load sample, the normalized sample features of that load sample are calculated in the clustering results. Z The functional expression for the total weighted distance on each feature is: ； in, For load samples m Total weighted distance M The total number of load samples, w z For the first z Weighting coefficients for each sample feature, Normalized load sample m The z Individual sample features, For load samples m The z Cluster centers corresponding to the features of each sample.

7. A system for analyzing the waveform characteristics and classifying the complexity of electrical equipment loads, comprising a microprocessor and a memory interconnected, characterized in that, The microprocessor is programmed or configured to execute the method for analyzing the load waveform characteristics and classifying the complexity of electrical equipment as described in any one of claims 1 to 6.

8. A computer-readable storage medium storing a computer program or instructions, characterized in that, The computer program or instructions are programmed or configured to execute, via a processor, the method for analyzing the load waveform characteristics and classifying the complexity of electrical equipment as described in any one of claims 1 to 6.

9. A computer program product, comprising a computer program or instructions, characterized in that, The computer program or instructions are programmed or configured to execute, via a processor, the method for analyzing the load waveform characteristics and classifying the complexity of electrical equipment as described in any one of claims 1 to 6.

Images