Power quality control method and device based on disturbance classification, equipment and medium

Abstract

The application discloses a power quality control method and device based on disturbance classification, equipment and medium. Among them, the method comprises the following steps: obtaining the original three-phase voltage data of the transformer, and performing a preprocessing operation on the original three-phase voltage data to obtain preprocessed three-phase voltage data; the voltage disturbance feature is extracted from the preprocessed three-phase voltage data; based on the voltage disturbance feature, the abnormality detection is carried out, and the voltage disturbance classification result is determined through the abnormality detection result and the voltage disturbance feature; the target device control strategy is determined based on the voltage disturbance classification result, so as to control the power quality through the target device control strategy. The technical scheme can determine the target device control strategy according to the analysis of the voltage disturbance feature, realize the accurate control of the power quality problem, and improve the accuracy of the power quality management.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power systems, and particularly relates to a power quality control method and device based on disturbance classification, equipment and medium. BACKGROUND

[0002] The distribution system is a new type of power system mainly constructed by new energy. The distribution transformer is the most critical electrical equipment in the distribution system, which has the functions of interconnection and power exchange of different voltage level power grids. Therefore, the safe and stable operation of the distribution transformer has a direct impact on the safety and stability of the distribution system. However, with the increasing proportion of large-scale distributed photovoltaic, electric vehicles and battery energy storage and other new energy in the distribution network, the application requirements of the distribution network in power quality control, power grid operation and regulation, source and load multi-element fusion and other aspects are increasing, which brings new challenges to the safe operation of the distribution network.

[0003] Although the traditional transformer has the advantages of low cost, high reliability and high efficiency, its function is too single and cannot meet the requirements of intelligent distribution network construction. However, the power electronic transformer with all power electronic devices has complex structure, many switching devices and high cost, and the control of power quality is a problem to be solved by those skilled in the art. SUMMARY

[0004] The present application provides a power quality control method and device based on disturbance classification, which can determine the target device control strategy according to the analysis of voltage disturbance characteristics, realize accurate control of power quality problems, and improve the accuracy of power quality management.

[0005] According to one aspect of the present application, a power quality control method based on disturbance classification is provided, comprising:

[0006] Obtaining original three-phase voltage data of a transformer, and performing a preprocessing operation on the original three-phase voltage data to obtain preprocessed three-phase voltage data;

[0007] Performing feature extraction on the preprocessed three-phase voltage data to obtain voltage disturbance characteristics;

[0008] Performing anomaly detection based on the voltage disturbance characteristics, and determining a voltage disturbance classification result through the anomaly detection result and the voltage disturbance characteristics;

[0009] Determining a target device control strategy based on the voltage disturbance classification result, so as to control power quality through the target device control strategy.

[0010] Optionally, the preprocessed three-phase voltage data includes instantaneous phasor and sequence component of three-phase voltage.

[0011] The three-phase voltage original data is acquired, and a preprocessing operation is performed on the three-phase voltage original data to obtain three-phase voltage data, including:

[0012] The three-phase voltage original data is acquired, and a filter is used to filter the three-phase voltage original data to obtain filtered three-phase voltage data.

[0013] The filtered three-phase voltage data is subjected to wavelet packet transform processing to obtain transformed three-phase voltage data.

[0014] The instantaneous phasor and sequence component of the three-phase voltage are determined according to the transformed three-phase voltage data.

[0015] Optionally, the voltage disturbance feature is extracted from the preprocessed three-phase voltage data, including:

[0016] The time-frequency analysis is performed on the preprocessed three-phase voltage data to obtain time-frequency analysis results.

[0017] The time-frequency feature matrix and texture feature are determined according to the time-frequency analysis results.

[0018] The instantaneous amplitude and phase of the three-phase voltage are determined according to the instantaneous phasor of the three-phase voltage.

[0019] The voltage disturbance feature is obtained by feature fusion based on the time-frequency feature matrix, texture feature, instantaneous amplitude, and phase.

[0020] Optionally, the time-frequency feature matrix and texture feature are determined according to the time-frequency analysis results, including:

[0021] The time-domain energy, frequency-domain energy, and time-frequency energy distribution data are determined according to the time-frequency analysis results.

[0022] The time-frequency feature matrix is obtained by combining the time-domain energy, frequency-domain energy, and time-frequency energy distribution data.

[0023] The texture feature is determined according to the time-frequency analysis results and texture analysis algorithm.

[0024] Optionally, the voltage disturbance feature is used for anomaly detection, and a voltage disturbance classification result is determined by the anomaly detection result and the voltage disturbance feature, including:

[0025] The voltage disturbance feature is input into a set data description model for anomaly detection to obtain an anomaly detection result.

[0026] If the anomaly degree in the anomaly detection result is greater than a set threshold, wavelet singularity detection is performed on the preprocessed three-phase voltage data to determine the starting time and ending time of the voltage disturbance.

[0027] constructing a multi-resolution phase space according to a starting time and an ending time of the voltage disturbance;

[0028] determining a dynamic invariant of the voltage signal by using the multi-resolution phase space, classifying the voltage disturbance based on the dynamic invariant to obtain a voltage disturbance classification result.

[0029] Optionally, the target device control strategy is determined based on the voltage disturbance detection result, and the target device control strategy comprises:

[0030] determining a state space model and a state observer;

[0031] determining an initial device control strategy based on the state space model and the state observer;

[0032] adjusting the initial device control strategy based on the voltage disturbance classification result to obtain the target device control strategy.

[0033] Optionally, after the power quality is controlled by using the target device control strategy, the method further comprises:

[0034] determining a system performance index according to a control result of controlling the power quality according to the target device control strategy;

[0035] performing performance comprehensive evaluation based on the system performance index to obtain a performance report.

[0036] According to another aspect of the present application, a power quality control device based on disturbance classification is provided, comprising:

[0037] a data acquisition module configured to acquire original three-phase voltage data of a transformer, and perform a preprocessing operation on the original three-phase voltage data to obtain preprocessed three-phase voltage data;

[0038] a feature extraction module configured to perform feature extraction on the preprocessed three-phase voltage data to obtain voltage disturbance features;

[0039] a voltage disturbance classification result determination module configured to perform abnormality detection based on the voltage disturbance features, and determine a voltage disturbance classification result based on an abnormality detection result and the voltage disturbance features;

[0040] a control strategy determination module configured to determine a target device control strategy based on the voltage disturbance classification result, so as to control power quality by using the target device control strategy.

[0041] Optionally, the preprocessed three-phase voltage data comprises instantaneous phasors and sequence components of three-phase voltage.

[0042] The data acquisition module is specifically configured to acquire three-phase voltage original data, perform filtering processing on the three-phase voltage original data by using a set filter to obtain filtered three-phase voltage data, perform wavelet packet transform processing on the filtered three-phase voltage data to obtain transformed three-phase voltage data, and determine instantaneous phasors and sequence components of the three-phase voltage according to the transformed three-phase voltage data.

[0043] Optionally, the feature extraction module comprises:

[0044] The data time-frequency analysis unit is configured to perform time-frequency analysis on the preprocessed three-phase voltage data to obtain a time-frequency analysis result.

[0045] The feature determination unit is configured to determine a time-frequency feature matrix and a texture feature according to the time-frequency analysis result.

[0046] The voltage instantaneous data determination unit is configured to determine instantaneous amplitude and phase of the three-phase voltage according to the instantaneous phasors of the three-phase voltage.

[0047] The feature fusion unit is configured to perform feature fusion based on the time-frequency feature matrix, the texture feature, the instantaneous amplitude, and the phase to obtain a voltage disturbance feature.

[0048] Optionally, the feature determination unit is specifically configured to determine time-domain energy, frequency-domain energy, and time-frequency energy distribution data according to the time-frequency analysis result, combine the time-domain energy, the frequency-domain energy, and the time-frequency energy distribution data to obtain a time-frequency feature matrix, and determine a texture feature according to the time-frequency analysis result and a texture analysis algorithm.

[0049] Optionally, the voltage disturbance classification result determination module is specifically configured to input the voltage disturbance feature into a set data description model to perform anomaly detection, obtain an anomaly detection result, perform wavelet singularity detection on the preprocessed three-phase voltage data if an abnormality degree in the anomaly detection result is greater than a set threshold, determine a starting time and an ending time of a voltage disturbance, construct a multi-resolution phase space according to the starting time and the ending time of the voltage disturbance, determine dynamic invariants of a voltage signal by using the multi-resolution phase space, and perform voltage disturbance classification based on the dynamic invariants to obtain a voltage disturbance classification result.

[0050] Optionally, the control strategy determination module is specifically configured to determine a state space model and a state observer, determine an initial device control strategy based on the state space model and the state observer, and adjust the initial device control strategy based on the voltage disturbance classification result to obtain a target device control strategy.

[0051] Optionally, the method further comprises:

[0052] The performance evaluation module is configured to determine a system performance index according to a control result of the target device control strategy in controlling power quality, and perform comprehensive performance evaluation based on the system performance index to obtain a performance report.

[0053] According to another aspect of the present application, there is provided an electronic device comprising:

[0054] at least one processor; and

[0055] a memory in communication with the at least one processor; wherein

[0056] The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor to enable the at least one processor to perform the power quality control method based on disturbance classification according to any one of the embodiments of the present application.

[0057] According to another aspect of the present application, there is provided a computer readable storage medium storing computer instructions for enabling a processor to perform the power quality control method based on disturbance classification according to any one of the embodiments of the present application when executed by the processor.

[0058] The technical solution of the embodiment of the present application comprises the following steps: obtaining original three-phase voltage data of a transformer, performing a pretreatment operation on the original three-phase voltage data to obtain pretreated three-phase voltage data, performing feature extraction on the pretreated three-phase voltage data to obtain voltage disturbance features, performing abnormality detection based on the voltage disturbance features, determining a voltage disturbance classification result through the abnormality detection result and the voltage disturbance features, and determining a target device control strategy based on the voltage disturbance classification result to control power quality through the target device control strategy. The technical solution can determine the target device control strategy according to the analysis of the voltage disturbance features, accurately control power quality, and improve the accuracy of power quality management.

[0059] It should be understood that the content described in this part is not intended to identify key or important features of the embodiments of the present application, nor is it used to limit the scope of the present application. Other features of the present application will become apparent from the following description. BRIEF DESCRIPTION OF DRAWINGS

[0060] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0061] Figure 1 is a flow chart of a power quality control method based on disturbance classification according to an embodiment of the present application;

[0062] Figure 2 is a flow chart of a power quality control method based on disturbance classification according to an embodiment of the present application;

[0063] Figure 3 is a structural schematic diagram of a power quality comprehensive management system according to an embodiment of the present application;

[0064] Figure 4 is a structural schematic diagram of a power quality control device based on disturbance classification according to an embodiment of the present application;

[0065] Figure 5 is a structural schematic diagram of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0066] In order to make the personnel in the art better understand the present application, the technical solutions in the embodiments of the present application will be described clearly and completely in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all the other embodiments obtained by the person of ordinary skill in the art without making creative efforts should belong to the scope of protection of the present application.

[0067] It should be noted that the terms "initial", "target" and the like in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and do not have to be used to describe a specific order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device including a series of steps or units does not have to be limited to those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0068] Embodiment one

[0069] Figure 1is a flow chart of a power quality control method based on disturbance classification according to an embodiment of the present application. The embodiment can be applicable to the case of controlling the power quality of a power distribution network. The method can be executed by a power quality control device based on disturbance classification. The power quality control device based on disturbance classification can be implemented in the form of hardware and / or software and can be configured in an electronic device with data processing capability. Figure 1 As shown in FIG. 1, the method comprises the following steps.

[0070] In S110, original three-phase voltage data of a transformer is acquired, and a pretreatment operation is performed on the original three-phase voltage data to obtain pretreated three-phase voltage data.

[0071] The transformer can be a three-phase three-winding transformer. The original three-phase voltage data can be sampling data of three-phase voltage collected by the transformer. In the embodiment, the original three-phase voltage data of the transformer can be acquired by sampling the three-phase voltage using a high-speed analog-to-digital converter, and the sampling frequency is set to 50 kHz. The sampling data is stored as an original voltage sampling data set, which contains three-phase voltage data. Each phase voltage data is composed of a series of sampling point values arranged in time sequence. The pretreatment operation can include adaptive filtering processing, wavelet packet transform processing, and threshold denoising processing. The pretreated three-phase voltage data can be data obtained by performing adaptive filtering processing and wavelet packet transform processing on the original three-phase voltage data. In the embodiment, the pretreated three-phase voltage data can include time-frequency representation, instantaneous phasor, and sequence component of the three-phase voltage.

[0072] In the embodiment, the original three-phase voltage data of the transformer can be filtered by an adaptive filter to obtain filtered three-phase voltage data. The filtered three-phase voltage data can be subjected to wavelet packet transform for time-frequency analysis. The wavelet packet transform result can be used for adaptive threshold denoising processing to obtain the pretreated three-phase voltage data.

[0073] In the embodiment, the pretreated three-phase voltage data includes instantaneous phasor and sequence component of the three-phase voltage. The original three-phase voltage data is acquired, and a pretreatment operation is performed on the original three-phase voltage data to obtain three-phase voltage data, including: acquiring the original three-phase voltage data, filtering the original three-phase voltage data using a set filter to obtain filtered three-phase voltage data; performing wavelet packet transform on the filtered three-phase voltage data to obtain transformed three-phase voltage data; and determining the instantaneous phasor and sequence component of the three-phase voltage based on the transformed three-phase voltage data.

[0074] The set filter can be a pre-set filter. In this embodiment, the set filter can be an adaptive Savitzky-Golay filter, and can also be set according to actual needs. The filtered three-phase voltage data can be data obtained after high-frequency noise removal by the adaptive Savitzky-Golay filter. In this embodiment, the original three-phase voltage data obtained by high-speed sampling is subjected to high-frequency noise removal by the improved adaptive Savitzky-Golay filter. In this embodiment, the filter order can be calculated according to the pre-estimated signal-to-noise ratio, the signal-to-noise ratio is divided by ten and rounded down, and then one is added. Then, the window length is calculated based on the pre-set fundamental frequency and the sampling frequency, the sampling frequency is divided by the fundamental frequency and rounded up, and then two is multiplied by one. The Savitzky-Golay filter is applied to each phase voltage data in the original voltage sampling data set by using the calculated filter order and window length, and the filtered three-phase voltage data is obtained. The application of the adaptive Savitzky-Golay filter can effectively remove high-frequency noise while retaining important features of the signal, especially when dealing with nonlinear and non-stationary power system signals.

[0075] The wavelet packet transform processing can be wavelet basis decomposition processing of the filtered three-phase voltage data. In this embodiment, the obtained filtered three-phase voltage data set is subjected to time-frequency analysis by applying wavelet packet transform. Specifically, Daubechies4 can be selected as the wavelet basis function, and the decomposition layer number can be obtained by calculating the logarithm to the base of two of the ratio of the sampling frequency to the fundamental frequency. In this embodiment, wavelet packet transform can be performed on each phase voltage data in the filtered three-phase voltage data to obtain the corresponding time-frequency representation. In this embodiment, the wavelet packet transform result obtained by wavelet packet transform processing can be subjected to adaptive threshold denoising by using the improved SURE (Stein's Unbiased Risk Estimate) threshold method. For each node of each decomposition layer, the threshold value is calculated, and the calculation method can be to multiply the noise standard deviation estimate of the node by the square root of the natural logarithm of two times the number of sampling points. In this embodiment, the calculated threshold value is used to perform soft threshold processing on the wavelet packet coefficients in the wavelet packet transform result to obtain denoised wavelet packet coefficients; the obtained denoised wavelet packet coefficients are used to reconstruct the voltage signal by inverse wavelet packet transform. The denoised wavelet packet coefficients of each phase voltage are subjected to inverse wavelet packet transform to obtain reconstructed high-quality three-phase voltage data, which includes the reconstructed voltage data of the three phases, i.e., the transformed three-phase voltage data.

[0076] For example, the adaptive threshold denoising process in this embodiment can be as follows: Obtain the output wavelet packet transform results WPT_a, WPT_b, and WPT_c. For each node k of each decomposition layer j, calculate the noise standard deviation estimate σj,k. Using the median absolute deviation method, take the median of the absolute values ​​of the wavelet coefficients and divide it by 0.6745 to obtain σj,k. Calculate the SURE threshold for each node using the calculated σj,k. Calculate the threshold using the formula λj,k=σj,k*sqrt(2*log(N)), where N is the number of sampling points. Store the calculated threshold in the threshold matrix Λ, with the same matrix dimension as the wavelet packet transform results. Apply the calculated SURE threshold to each coefficient in the obtained WPT_a, WPT_b, and WPT_c for soft thresholding. If the absolute value of the coefficient is less than the corresponding threshold, set it to zero; otherwise, retain the sign of the difference between it and the threshold. Obtain the denoised wavelet packet coefficients WPT_denoised_a, WPT_denoised_b, and WPT_denoised_c. Calculate the signal-to-noise ratio (SNR) improvement before and after denoising. Calculate the signal energy and noise energy using the original wavelet packet coefficients and the denoised coefficients, respectively. Calculate the SNR before and after denoising to obtain the SNR improvement value ΔSNR. Compare ΔSNR with a preset threshold. If it is lower than the threshold, return to the step of recalculating the SURE threshold for each node and adjust the threshold calculation parameters; otherwise, output the final denoising result.

[0077] In this context, the instantaneous phasor of three-phase voltage refers to the complex representation of the voltage of each phase in a three-phase AC system at a specific moment. The sequence component of three-phase voltage refers to the decomposition of an unbalanced three-phase voltage into symmetrical components, namely, positive-sequence, negative-sequence, and zero-sequence components. In this embodiment, the specific method for determining the instantaneous phasor and sequence component of three-phase voltage based on the transformed three-phase voltage data is as follows: The Hilbert transform is applied to each phase voltage data after reconstruction to obtain the corresponding analytic signal. The analytic signal is represented in complex form to obtain the instantaneous phasor of the three phases. Then, the symmetrical component method is used to calculate the instantaneous positive-sequence component, instantaneous negative-sequence component, and instantaneous zero-sequence component. The instantaneous positive-sequence component is calculated by adding the instantaneous phasor of the first phase multiplied by the complex root of unity, adding the instantaneous phasor of the third phase multiplied by the complex root of unity, and finally dividing the result by three. The instantaneous negative sequence component is calculated by adding the first-phase instantaneous phasor to the second-phase instantaneous phasor multiplied by the square of the complex root of unity, then adding the third-phase instantaneous phasor multiplied by the first power of the complex root of unity, and finally dividing the result by three. The instantaneous zero sequence component is calculated by adding the three-phase instantaneous phasors and dividing by three. Here, the complex root of unity is e, which is the exponent of the product of 2π divided by three and the imaginary unit.

[0078] Through the setting in this embodiment, high-quality three-phase voltage data acquisition and preprocessing can be realized through high-speed sampling, adaptive filtering, wavelet packet transformation and other processing operations, and the accuracy and real-time performance of power quality monitoring are improved, thereby providing high-quality data support for comprehensive power quality management.

[0079] S120, performing feature extraction on the preprocessed three-phase voltage data to obtain voltage disturbance features.

[0080] The voltage disturbance features can be feature vectors obtained by extracting time domain, frequency domain and time-frequency domain characteristics of the voltage signal. In this embodiment, voltage disturbance feature extraction operations can be performed on the preprocessed three-phase voltage data, time-frequency representation, instantaneous phasor and sequence component. In this embodiment, Stockwell transform can be performed on each phase voltage data in the preprocessed three-phase voltage data, and Stockwell transform results can be obtained. Time-frequency feature matrix and texture features can be extracted from the Stockwell transform results, and fusion processing is performed to obtain voltage disturbance features.

[0081] S130, performing anomaly detection based on the voltage disturbance features, and determining a voltage disturbance classification result through the anomaly detection result and the voltage disturbance features.

[0082] The anomaly detection result can be an anomaly degree index obtained by inputting the voltage disturbance features into a trained model for anomaly detection. In this embodiment, the voltage disturbance features can be input into a trained support vector data description model for anomaly detection, and a quantitative index of the anomaly degree can be obtained. In this embodiment, the quantitative index of the anomaly degree can be used to accurately identify and quantitatively describe the voltage disturbance features through wavelet singularity detection, multi-resolution phase space reconstruction, dynamic invariant calculation, empirical mode decomposition, Hilbert-Huang transform and fuzzy reasoning, thereby obtaining the final voltage disturbance classification result.

[0083] S140, determining a target device control strategy based on the voltage disturbance classification result, to control power quality through the target device control strategy.

[0084] The target device control strategy can be understood as a specific way of controlling the switching device of the transformer. In this embodiment, a state space model can be constructed based on the voltage disturbance classification result, a robust observer can be designed, model predictive control can be applied, control weights can be dynamically adjusted, and the way of solving optimization problems can be obtained to generate PWM control signals. The switching of the power electronic device is controlled according to the PWM control signal as the target device control strategy, to realize accurate control of power quality.

[0085] The technical scheme of the embodiment of the present application obtains original three-phase voltage data of a transformer, performs preprocessing operation on the original three-phase voltage data to obtain preprocessed three-phase voltage data, performs feature extraction on the preprocessed three-phase voltage data to obtain voltage disturbance features, performs abnormality detection based on the voltage disturbance features, determines a voltage disturbance classification result through the abnormality detection result and the voltage disturbance features, determines a target device control strategy based on the voltage disturbance classification result, and controls power quality through the target device control strategy. The technical scheme can determine the target device control strategy according to the analysis of the voltage disturbance features, realizes accurate control of power quality problems, and improves the accuracy of power quality management.

[0086] Embodiment two

[0087] Figure 2 is a flowchart of a power quality control method based on disturbance classification according to the second embodiment of the present application, which is optimized based on the above-mentioned embodiment. The specific optimization is that the voltage disturbance features are obtained by performing feature extraction on the preprocessed three-phase voltage data, including: performing time-frequency analysis on the preprocessed three-phase voltage data to obtain time-frequency analysis results; determining a time-frequency feature matrix and a texture feature according to the time-frequency analysis results; determining an instantaneous amplitude and a phase of the three-phase voltage according to instantaneous phasors of the three-phase voltage; and performing feature fusion based on the time-frequency feature matrix, the texture feature, the instantaneous amplitude, and the phase to obtain the voltage disturbance features. As shown in Figure 2 , the method includes:

[0088] S210, original three-phase voltage data of a transformer is obtained, and preprocessing operation is performed on the original three-phase voltage data to obtain preprocessed three-phase voltage data.

[0089] S220, time-frequency analysis is performed on the preprocessed three-phase voltage data to obtain time-frequency analysis results.

[0090] The time-frequency analysis can be a localized analysis of the preprocessed three-phase voltage data at different times and frequencies by a time-frequency analysis method. The time-frequency analysis method in this embodiment can be a Stockwell transform method, which is a time-frequency analysis method that extends the short-time Fourier transform (STFT) and the wavelet transform (WT). In this embodiment, the preprocessed three-phase voltage data can be subjected to Stockwell transform for time-frequency analysis to obtain the corresponding video analysis result. In this embodiment, the Stockwell transform can be performed on each phase voltage data, specifically, the product of the voltage data and a Gaussian window function can be subjected to Fourier transform. The width of the Gaussian window function varies with the frequency, and the time-frequency resolution is adaptively adjusted. In this embodiment, the Stockwell transform is performed on each phase voltage data to obtain the Stockwell transform result of each phase voltage data, i.e., the time-frequency analysis result, which can be represented as three-dimensional data of time, frequency, and amplitude.

[0091] In S230, a time-frequency feature matrix and a texture feature are determined according to the time-frequency analysis result.

[0092] The time-frequency feature matrix can be extracted from the time-frequency analysis result. In this embodiment, the time-frequency feature matrix can be extracted from the Stockwell transform result of each phase voltage data. The time-frequency feature matrix can include a feature matrix obtained by combining time-domain energy, frequency-domain energy, and time-frequency energy distribution data. The texture feature can be a texture feature extracted by a texture analysis algorithm from the time-frequency analysis result. In this embodiment, the texture feature of the three-phase voltage data can be extracted by applying an improved local binary pattern algorithm to the Stockwell transform result of each phase voltage data.

[0093] In this embodiment, the time-frequency feature matrix and the texture feature determined according to the time-frequency analysis result can include: determining time-domain energy, frequency-domain energy, and time-frequency energy distribution data according to the time-frequency analysis result; combining the time-domain energy, the frequency-domain energy, and the time-frequency energy distribution data to obtain the time-frequency feature matrix; and determining the texture feature according to the time-frequency analysis result and a texture analysis algorithm.

[0094] The time-frequency energy can be the energy distribution of the three-phase voltage data in the time dimension. The frequency-domain energy can be the energy distribution of the three-phase voltage data in the frequency dimension. The time-frequency energy distribution data can refer to the position of the energy distribution, the dispersion degree of the distribution, and the skewness and kurtosis of the distribution. The time-frequency energy distribution data in this embodiment can include time-frequency centroid, time-frequency variance, and skewness and kurtosis of the time-frequency distribution.

[0095] In this embodiment, the Stockwell transform result in the time-frequency analysis result can be used to extract the time-frequency feature matrix. Specifically, time domain energy and frequency domain energy are calculated, which correspond to the integration of the Stockwell transform result in the time dimension and the frequency dimension, respectively; time-frequency centroid is calculated, which represents the center position of the energy distribution; time-frequency variance is calculated, which represents the dispersion degree of the energy distribution; skewness and kurtosis of the time-frequency distribution are calculated, which represent the asymmetry and sharpness of the distribution, respectively. In this embodiment, the specific way of obtaining the time-frequency feature matrix according to the time domain energy, the frequency domain energy, and the time-frequency energy distribution data can be to combine the time domain energy, the frequency domain energy, the time-frequency centroid, the time-frequency variance, and the skewness and kurtosis of the time-frequency distribution into the time-frequency feature matrix.

[0096] For example, the specific way of determining the time-frequency feature matrix according to the time-frequency analysis result in this embodiment can be: obtaining the Stockwell transform results S_a(τ, f), S_b(τ, f), and S_c(τ, f) in the time-frequency analysis result. Calculating the time domain energy E_t. For each frequency f, the integral of |S(τ, f)|^2 over time τ is calculated. The time domain energy E_t_a, E_t_b, and E_t_c of the three phases are obtained. The Stockwell transform result of each phase voltage data is used to calculate the frequency domain energy E_f. For each time τ, the integral of |S(τ, f)|^2 over frequency f is calculated. The frequency domain energy E_f_a, E_f_b, and E_f_c of the three phases are obtained. The time-frequency centroid is calculated according to the time domain energy of the three phases and the frequency domain energy of the three phases. The time centroid Centroid_t is the weighted average of |S(τ, f)|^2 and τ, and the frequency centroid Centroid_f is the weighted average of |S(τ, f)|^2 and f. The time-frequency centroid Centroid_t_abc and Centroid_f_abc of the three phases are obtained. The time-frequency variance is calculated using the Stockwell transform result in the time-frequency analysis result. The time variance Var_t is the weighted average of |S(τ, f)|^2 and (τ-Centroid_t)^2, and the frequency variance Var_f is the weighted average of |S(τ, f)|^2 and (f-Centroid_f)^2. The time-frequency variance Var_t_abc and Var_f_abc of the three phases are obtained. The skewness and kurtosis of the time-frequency distribution are calculated using the Stockwell transform result in the time-frequency analysis result. The skewness is the third central moment divided by the square of the variance, and the kurtosis is the fourth central moment divided by the square of the variance. The skewness Skewness_abc and the kurtosis Kurtosis_abc of the three phases are obtained. The calculated features are combined into the time-frequency feature matrix TF_features. Each row of the matrix corresponds to a time-frequency feature, and each column corresponds to one phase of the three phases. The output TF_features is used for subsequent analysis.

[0097] The texture analysis algorithm can be used for image local feature comparison and texture analysis, so as to extract local texture features of the image. The texture analysis algorithm in the embodiment can be a local binary pattern algorithm, and can also be other texture analysis algorithms, which can be set according to actual needs.

[0098] The time-frequency analysis result in the embodiment can include a Stockwell transform amplitude graph. The Stockwell transform amplitude graph can be analyzed and processed by using a local binary pattern algorithm to extract texture features. Specifically, a rotation-invariant uniform local binary pattern operator can be used in the embodiment to process the Stockwell transform amplitude graph of each phase voltage. A binary code is generated by comparing the relationship between a center pixel and its neighborhood pixels, and a local binary pattern feature vector is formed by counting the distribution of the codes, that is, the texture features are obtained.

[0099] Through such a setting, the construction of the time-frequency feature matrix not only includes traditional energy distribution information, but also introduces high-order statistical features such as time-frequency centroid, variance, skewness and kurtosis, which greatly enhances the description ability of complex nonlinear disturbances. The improved local binary pattern algorithm considers the rotation invariance when extracting the texture features of the time-frequency graph, which makes the feature extraction more robust to phase changes of the signal, and is conducive to accurately identifying various types of voltage fluctuations.

[0100] S240, determining the instantaneous amplitude and phase of the three-phase voltage according to the instantaneous phasor of the three-phase voltage.

[0101] The instantaneous amplitude of the three-phase voltage, also known as the instantaneous value, can refer to the value of the voltage at a certain time. The phase of the three-phase voltage can be used to represent the position of the voltage waveform at a certain time. The instantaneous amplitude and phase of the three-phase voltage can be calculated by using the preprocessed instantaneous phasor of the three-phase voltage in the embodiment. Specifically, the modulus of each phase instantaneous phasor can be calculated to obtain the corresponding instantaneous amplitude, and the argument of each phase instantaneous phasor can be calculated to obtain the corresponding instantaneous phase. The calculated instantaneous amplitude and phase data can be stored separately for subsequent analysis in the embodiment.

[0102] S250, performing feature fusion to obtain voltage disturbance features based on the time-frequency feature matrix, the texture features, the instantaneous amplitude and the phase.

[0103] The feature fusion can be a splicing operation according to the feature order to fuse into a complete feature vector.

[0104] The statistical characteristics of the instantaneous amplitude and phase (including mean, standard deviation, maximum, minimum, skewness, kurtosis), the time-frequency feature matrix, and the texture feature vector are spliced in sequence to form a high-dimensional feature vector, i.e., the voltage disturbance feature. The feature vector comprehensively represents the time domain, frequency domain, and time-frequency domain characteristics of the voltage signal.

[0105] In S260, abnormality detection is performed based on the voltage disturbance feature, and a voltage disturbance classification result is determined based on the abnormality detection result and the voltage disturbance feature.

[0106] In the embodiment, the abnormality detection based on the voltage disturbance feature and the determination of the voltage disturbance classification result based on the abnormality detection result and the voltage disturbance feature are optional, and include: inputting the voltage disturbance feature into a set data description model to perform abnormality detection and obtain an abnormality detection result; if the abnormality degree in the abnormality detection result is greater than a set threshold, performing wavelet singularity detection on the preprocessed three-phase voltage data to determine the start time and the end time of the voltage disturbance; constructing a multi-resolution phase space according to the start time and the end time of the voltage disturbance; determining dynamic invariants of the voltage signal by using the multi-resolution phase space, and classifying the voltage disturbance based on the dynamic invariants to obtain the voltage disturbance classification result.

[0107] The set data description model can be a support vector data description model. Support vector data description (SVDD) is an important data description method that can be used for heterogeneous point detection or classification. The abnormality detection result can be a quantitative index of the abnormality degree. In the embodiment, a pre-trained support vector data description model can be used to perform abnormality detection on the voltage disturbance feature vector. Specifically, in the embodiment, the voltage disturbance feature vector can be input into the support vector data description model, the distance of the vector to the decision boundary can be calculated, and the distance can be taken as the quantitative index of the abnormality degree.

[0108] Exemplarily, the specific manner of the support vector data description anomaly detection supported in the embodiment can be to obtain a voltage disturbance feature vector F. F is standardized so that the mean of each feature is 0 and the standard deviation is 1. A standardized feature vector F norm is obtained. The pre-trained SVDD model parameters are read, including support vectors SV, Lagrange multipliers a and kernel function parameters g. These parameters are stored in a model parameter file. The kernel distance of F norm and each support vector SV i is calculated. A Gaussian kernel function K(F norm, SV i) = exp(-g||F norm-SV i||^2) is used. A kernel distance vector K dist is obtained. Using K dist and a, the decision function value f(F norm) = åa i *K(F norm, SV i)-p is calculated, where p is the threshold parameter of the SVDD model. A decision function value f value is obtained. The obtained f value is converted into an abnormality degree. The abnormality degree score = max(0, -f value). If the score is greater than a preset threshold threshold, it is marked as abnormal and triggers subsequent detailed analysis; otherwise, it is marked as normal; finally, the anomaly detection result is output, including the abnormality degree score and the abnormality flag flag. These results will be used for subsequent disturbance analysis and control strategy adjustment.

[0109] wherein the threshold value can be a pre-set numerical value, which can be set according to actual needs in the embodiment. Wavelet singularity detection is a method for detecting singular points in signals or data using wavelet transform theory. The start time and end time of the voltage disturbance correspond to the singular points obtained by wavelet singularity detection.

[0110] In the embodiment, if the abnormality degree in the anomaly detection result is greater than the pre-set threshold value, wavelet singularity detection is performed on the pre-processed three-phase voltage data, and the start time and end time of the voltage disturbance are determined according to the identified singular points. In the embodiment, the wavelet singularity detection can be performed on the wavelet packet transform result of the pre-processed three-phase voltage data. The modulus maximum points of the wavelet coefficients at each scale can be found, i.e. the positions where the absolute values of the coefficients are greater than those of the adjacent points. Then, the propagation characteristics of these modulus maximum points are compared across scales to identify the singular points in the signal. Since these singular points usually correspond to the start and end time of the voltage disturbance, the start time and end time of the point cloud disturbance can be determined through these singular points.

[0111] The multi-resolution phase space can refer to a space for representing and analyzing the phase and amplitude of a signal at different resolutions. In this embodiment, the amplitude and phase changes of the three-phase voltage data can be observed and analyzed at different scales. In this embodiment, the multi-resolution phase space can be constructed by the preprocessed three-phase voltage data and the singular points identified by the wavelet singularity detection. Specifically, three different time delays are selected to construct three phase spaces with the same embedding dimension. Each phase space is composed of the original voltage data and its different delayed versions. This multi-resolution reconstruction can capture the dynamic characteristics of the signal at different time scales.

[0112] The specific way of calculating the dynamic invariants of the voltage signal using the constructed multi-resolution phase space in this embodiment can be as follows: first, calculate the correlation dimension by estimating the limit of the ratio of the logarithm of the correlation integral to the logarithm of the scale at different scales; then, calculate the maximum Lyapunov exponent by tracking the separation rate of nearby trajectories in the phase space; finally, calculate the approximate entropy by comparing the differences in pattern matching probability at different embedding dimensions. These dynamic invariants represent the complexity and prediction difficulty of the voltage signal.

[0113] For example, the reconstructed trajectories at different embedding dimensions m and time delays τ can be calculated using the constructed multi-resolution phase spaces X(t), Y(t), and Z(t) in this embodiment. For each reconstructed trajectory, the Euclidean distance between pairs of points is calculated to obtain the distance matrix D_m. Based on the distance matrix D_m, the correlation integral C(ε) is calculated. For different distance thresholds ε, the proportion of distance logs less than ε in the total distance logs is counted. The correlation integral curve C(ε) versus ε is obtained. Using the correlation integral curve, the correlation dimension D is estimated. In the double logarithmic coordinates, the linear relationship of log(C(ε)) and log(ε) is fitted. The slope of the linear part is the correlation dimension D. The estimated correlation dimension D_est is output. Based on the phase space reconstruction data of the multi-resolution phase space, the maximum Lyapunov exponent λ is calculated. A reference trajectory is selected in the phase space, and the separation process of the adjacent trajectory is tracked. The change of the separation distance with time is fitted, and the slope is the maximum Lyapunov exponent λ. The maximum Lyapunov exponent λ_max is output. Using the reconstruction data of the multi-resolution phase space, the approximate entropy ApEn is calculated. For different pattern lengths m, the pattern matching probability is calculated. The difference in pattern matching probability between m and m+1 lengths is calculated to obtain the approximate entropy ApEn value. The estimated correlation dimension D_est, the output maximum Lyapunov exponent λ_max, and the approximate entropy ApEn value are integrated to form a dynamic invariant vector DI = [D_est, λ_max, ApEn], and the DI is output for subsequent disturbance characteristic analysis and classification.

[0114] Further, in this embodiment, the improved empirical mode decomposition algorithm is also applied to the pre-processed three-phase voltage data. Specifically, through an iterative screening process, the original signal is decomposed into a series of intrinsic mode functions and a residual term. Each intrinsic mode function represents an inherent oscillation mode of the signal. The Hilbert transform is calculated for each intrinsic mode function to obtain the analytic signal. The instantaneous frequency and amplitude are extracted from the analytic signal to form the Hilbert spectrum. Based on the obtained intrinsic mode functions and their Hilbert spectrum, the Hilbert-Huang spectrum is constructed. The instantaneous amplitude and frequency of each intrinsic mode function are represented in the time-frequency plane. The time-frequency representation of all intrinsic mode functions is superimposed to obtain the complete Hilbert-Huang spectrum. This spectrum clearly shows the time-varying frequency characteristics of the signal, which helps to identify complex nonlinear and non-stationary disturbances.

[0115] In this embodiment, based on the determination of dynamic invariants, singular point detection, and the Hilbert-Huang spectrum of voltage signals, a Takagi-Sugeno fuzzy reasoning system is constructed for voltage disturbance classification to obtain the voltage disturbance classification results. Specifically, first, define fuzzy sets and membership functions for each input feature; then, design a series of if-then rules, where the antecedent part of each rule is a fuzzy description of the input features, and the consequent part is a linear function; use the input features to activate these rules and calculate the trigger strength of each rule. Finally, the final disturbance classification results and confidence are obtained by weighted averaging.

[0116] Through such a setting in this embodiment, the depth and breadth of power quality analysis are greatly improved, providing strong theoretical support and data foundation for formulating precise compensation strategies and preventive maintenance plans. In this embodiment, through techniques such as wavelet singularity detection, multi-resolution phase space reconstruction, dynamic invariant calculation, empirical mode decomposition, and Takagi-Sugeno fuzzy reasoning system, accurate identification and quantitative analysis of voltage disturbances are achieved. In the specific scenario of comprehensive power quality management, they can accurately identify and quantitatively describe various complex voltage disturbances, including harmonics, interharmonics, voltage flicker, transient overvoltage, etc., improving the system's diagnostic ability for various power quality problems. Secondly, by calculating dynamic invariants, the system can assess the overall stability of the power grid and predict potential voltage collapse risks, which is of great significance for the safe operation of large power systems. Furthermore, the combination of EEMD and Hilbert-Huang transform enables the system to accurately separate and quantitatively analyze complex power quality problems, such as the superposition of harmonics and flicker, the coexistence of transient and persistent disturbances, etc. Finally, the application of fuzzy reasoning systems improves the interpretability and reliability of classification results, especially in dealing with boundary fuzzy disturbance types, it can give more reasonable classification results and confidence.

[0117] S270, determining a target device control strategy based on the voltage disturbance classification result, so as to control power quality through the target device control strategy.

[0118] In this embodiment, the target device control strategy is determined based on the voltage disturbance detection result, including: determining a state space model and a state observer; determining an initial device control strategy based on the state space model and the state observer; and adjusting the initial device control strategy based on the voltage disturbance classification result to obtain the target device control strategy.

[0119] The state space model is a dynamic time-domain model, which takes implicit time as the independent variable and describes the dynamic behavior of the system through state equations and output equations. In this embodiment, the state space model of the control system can be constructed based on the voltage disturbance detection result, and the results and data obtained after various analyses of the preprocessed three-phase voltage. The specific way to determine the state space model in this embodiment is to define a state vector containing key variables such as voltage and current; define an input vector containing control variables and measurable disturbances; define an output vector containing variables to be controlled; and establish state equations and output equations according to the physical characteristics and dynamic behavior of the voltage data. The state equation describes the evolution of the system state, and the output equation describes the relationship between the system output and the state and input.

[0120] The state observer can be a kind of dynamic system that obtains an estimated value of a state variable according to the measured value of the external variables (input variables and output variables) of the system. In this embodiment, a robust H∞ state observer can be designed based on the constructed state space model. First, define the constraint set of system disturbances and measurement noise; then, construct a Riccati equation containing the system matrix and the noise covariance matrix; solve the Riccati equation to obtain a positive definite matrix solution. Use this matrix solution to calculate the observer gain matrix. Finally, construct the state observer equation, including the state prediction part and the error correction part.

[0121] For example, the robust H∞ state observer design procedure can be as follows. Obtain the constructed state-space model parameters, including system matrix A, input matrix B, output matrix C, and direct transfer matrix D. Define the system disturbance matrix E and the measurement noise matrix F. Set the H∞ performance index γ. Construct the generalized control system, taking the state estimation error as the controlled output and the system disturbance and measurement noise as the external input. Form the augmented system matrix [A, E; C, F]. Construct the algebraic Riccati equation AX + XAT + X(γ-2*EE^T - C^TC)X + BB^T = 0. Solve this equation using numerical methods (such as the Schur decomposition method) to obtain a positive definite matrix solution X. Using the obtained positive definite matrix solution X, calculate the observer gain matrix L = XC^T*R^(-1), where R is the measurement noise covariance matrix. Store L for subsequent state estimation. Construct the state observer equation dx^ / dt = Ax^ + Bu + L(y - Cx^- Du). Where x^ is the state estimation value, u is the control input, and y is the system measurement output. Perform observer stability analysis. Calculate the eigenvalues of A - LC, ensuring that the real parts of all eigenvalues are negative to guarantee the asymptotic stability of the observer. Output the stability analysis results.

[0122] The initial device control strategy can be an initial device control mode predicted according to a model. In this embodiment, a state space model and a state observer of the state space model are used to generate the initial device control strategy by model predictive control. A prediction horizon and a control horizon are defined. An objective function is constructed, including a state tracking error term and a control input penalty term. Constraint conditions are set, including state constraints and control input constraints. The problem is converted into a finite horizon optimization problem, and the objective is to minimize the objective function under the condition of meeting the constraints, to obtain the preliminary device control strategy. Then, the weight matrix of the model predictive control can be dynamically adjusted according to the voltage disturbance analysis result. The weight matrix of the state tracking error is adjusted according to the importance of different state variables and the current disturbance characteristics. The penalty weight matrix of the control input is adjusted according to the current system state and the control target. This adaptive weight adjustment can make the control strategy better cope with different types and degrees of disturbance. An improved interior point method can be used to solve the constructed model predictive control optimization problem. First, the constrained problem is converted into an unconstrained problem, and an obstacle function is introduced. Then, the modified KKT (Karush-Kuhn-Tucker) condition is solved by using the Newton method. Through the iterative process, the obstacle parameter is gradually reduced, and the original variable and the dual variable are updated. Finally, the optimal control sequence that meets the accuracy requirement is obtained. The obtained optimal control sequence is converted into a PWM (pulse width modulation) control signal. The space vector pulse width modulation algorithm is used to map the continuous control quantity to the discrete switching state. First, the reference voltage vector is calculated according to the control quantity. Then, the adjacent basic voltage vector is selected. The action time of each basic vector is calculated. Finally, according to the time sequence obtained by calculation, the specific PWM waveform is generated, and the switching of the power electronic device is controlled, that is, the target device control strategy is obtained.

[0123] An exemplary specific procedure for model predictive control application in this embodiment can be to obtain the state space model and state estimation x^ of the state observer in the step. Define the prediction horizon Np and control horizon Nc. Initialize the state prediction sequence X and control sequence U. Construct the prediction model: predict the states Np steps into the future using the state equation x(k+1) = Ax(k) + Bu(k). Form the state prediction matrix X = f(x^, U), where f represents the iterative prediction process. Define the objective function J = Σ(||x(k+i|k) - xref||^2_Q + ||u(k+i|k)||^2_R). Where Q is the state weight matrix, R is the control weight matrix, and xref is the reference trajectory. Express the objective function as a quadratic form in the control sequence U. Set the constraints. Include the control input constraints um in ≤ u(k+i|k) ≤ umax and state constraints xmin ≤ x(k+i|k) ≤ xmax. Express the constraints as linear inequalities in U, AU ≤ b. Combine the defined objective function and set constraints to form a standard quadratic programming problem: min 0.5 U^TH U + f^TU, s.t. AU ≤ b. Where H is the Hessian matrix of the objective function, and f is the first order term coefficient. Solve the optimization problem constructed by S435 using a quadratic programming solver (such as the interior point method). Obtain the optimal control sequence U*. Extract the first element of U, u(k|k), as the current control input.

[0124] Through such a setting in the embodiment, the accuracy, speed and adaptability of power quality treatment are significantly improved, providing strong technical support for achieving high-quality and reliable power supply. The embodiment realizes accurate control and compensation of power quality problems through the construction of an accurate state space model, the design of a robust H∞ state observer, the application of model predictive control, and the dynamic adjustment of control weights. In the specific scene of comprehensive treatment of power quality, it can achieve fast and accurate compensation for various voltage disturbances. For example, in the case of voltage sag, the system can respond quickly within a few cycles to restore the voltage to the allowable range, effectively protecting sensitive loads. For harmonic problems, the control strategy can accurately generate anti-phase compensation current, significantly reducing the total harmonic distortion (THD) and improving power quality. Secondly, the application of MPC enables the system to seek the best balance between multiple objectives. For example, when dealing with voltage imbalance problems, the controller can consider both negative sequence current compensation and power loss minimization to achieve optimal compensation effect. Furthermore, the use of a robust observer greatly improves the stability and reliability of the system in complex grid environments. Even in the case of rapid load changes or uncertain grid parameters, the system can maintain good control performance. This is particularly important for modern power grids that contain a large number of distributed energy sources and nonlinear loads. Finally, the dynamic weight adjustment strategy enables the system to flexibly respond to different types and degrees of disturbances. For example, when facing severe voltage sag, the system will automatically increase the weight of voltage stability control; while dealing with minor harmonic problems, it will focus more on energy efficiency. This adaptive capability ensures that the system maintains optimal performance under various operating conditions.

[0125] In the embodiment, after the power quality is controlled by the target device control strategy, the system performance index is determined according to the control result of the target device control strategy controlling the power quality; and the performance is comprehensively evaluated based on the system performance index to obtain a performance report.

[0126] The system performance index can be a performance index of the power quality control system, and can be used to evaluate the performance state of the entire power quality control system. In the embodiment, the system performance index can include voltage deviation, total harmonic distortion, voltage imbalance degree, and dynamic response time, etc. In the embodiment, the system performance can be evaluated and self-optimized based on the results of the above analysis and processing steps. The system can be continuously improved by calculating the performance index, applying fuzzy comprehensive evaluation, updating the optimization target, performing parameter optimization, and generating a performance report. The performance comprehensive evaluation can be a comprehensive evaluation result of the system performance obtained by establishing evaluation levels and membership functions through the fuzzy comprehensive evaluation method. The performance report can be a report formed based on the data in the performance comprehensive evaluation process of determining the system performance index.

[0127] The control results of the pre-processed voltage data and the target device control strategy can be used to calculate system performance indicators in this embodiment. Specifically, the voltage deviation, which is the relative error between the actual voltage and the reference voltage, can be calculated. The total harmonic distortion, which is the ratio of the sum of the squares of the harmonic voltage effective values to the fundamental voltage effective value, can be calculated. The voltage unbalance degree, which is the ratio of the negative sequence voltage component to the positive sequence voltage component, can be calculated. The dynamic response time, which is the time interval from the disturbance to the voltage recovery to the steady state, can be calculated. Based on the calculated performance indicators, a fuzzy comprehensive evaluation method is applied. First, evaluation levels and membership functions are established for each performance indicator. Then, a fuzzy relationship matrix is constructed, whose elements represent the membership of each indicator to each evaluation level. A weight vector is defined to reflect the relative importance of each indicator. Finally, through fuzzy synthesis operation, the comprehensive evaluation result of system performance is obtained. According to the comprehensive performance evaluation result, the objective function of parameter optimization is updated. The indicators such as voltage deviation, total harmonic distortion, voltage unbalance degree and dynamic response time are integrated into the objective function. According to the current system state and control demand, the weight of each indicator in the objective function is dynamically adjusted. This updated objective function will guide the subsequent parameter optimization process. Then, using the improved particle swarm optimization algorithm, the control parameters are optimized based on the updated objective function. Initialize the particle swarm, each particle represents a set of candidate control parameters. Define the position update formula and velocity update formula of the particle, which contain inertia weight, individual learning factor and social learning factor. Iteratively update the particle position and velocity, and evaluate the fitness of each particle. Record the global optimal solution and individual optimal solution until the termination condition is met. Apply the optimized parameters to the target device control strategy. Update the prediction model parameters used in model predictive control, including state transition matrix and input matrix. Update the gain matrix of the state observer to improve the accuracy of state estimation. Adjust the weight matrix of the objective function in model predictive control, including the weight of state tracking error and the penalty weight of control input. These updates will improve the performance and robustness of the control system. Based on the system evaluation process and the results, a detailed system performance report is generated. Summarize the current system state, including key electrical parameters and operation indicators. Analyze each performance indicator in detail, including its value, trend and deviation from the target value. List the optimized control parameters and compare them with the previous parameters. Provide optimization suggestions, including possible hardware upgrades and control strategy adjustments. Analyze historical data, draw trend charts of key indicators, and identify long-term performance changes and potential problems. Finally, a comprehensive evaluation report is generated, summarizing the overall performance of the system and proposing further improvement directions.

[0128] For example, the specific process of the fuzzy comprehensive evaluation based on the system performance indicators in the embodiment can be: obtaining the calculated performance indicators, including voltage deviation ΔV, total harmonic distortion THD, voltage unbalance degree VUF, and dynamic response time Tr. Define the evaluation level set V = {very good, good, general, poor, very poor}. Construct the membership function for each performance indicator. Use trapezoidal or Gaussian function to define the membership of each level. For example, define μΔV(x) = {μ1(x), μ2(x), μ3(x), μ4(x), μ5(x)} for ΔV, where μi(x) represents the membership of the ith level. Using the defined membership function, calculate the membership of each performance indicator to each evaluation level. Form the fuzzy relation matrix R, where Rij represents the membership of the ith indicator to the jth evaluation level. Determine the weight vector W = {w1, w2, w3, w4}, which represents the relative importance of each performance indicator. The weight can be determined using the analytic hierarchy process or expert scoring method. Ensure that the sum of the weights is 1. Perform fuzzy synthesis operation S = W○R, where○represents the fuzzy synthesis operator. Common synthesis operators include M(∧, ∨) and M(·, ∨). Get the comprehensive evaluation result vector S. Defuzzify the obtained evaluation result S. The maximum membership degree method or barycenter method can be used. Get the quantitative score and qualitative level of the system performance, and output the evaluation result for subsequent optimization decision.

[0129] The improved particle swarm optimization algorithm in this embodiment is implemented as follows: obtain the comprehensive performance evaluation result, and update the objective function J opt of parameter optimization. Define the position vector x of a particle to represent the control parameter to be optimized, and the velocity vector v to represent the update direction and step size of the parameter. Initialize the particle swarm, and randomly generate the position x and velocity v of N particles. Evaluate the fitness of each particle. Substitute the position x of each particle into the objective function J opt, and calculate the function value. Record the individual optimal position pi and the global optimal position pg of each particle. Update the velocity of each particle using the formula v(t+1) = wv(t) + c1r1*(pi-x(t)) + c2r2(pg-x(t)). Where w is the inertia weight, c1 and c2 are acceleration constants, and r1 and r2 are random numbers between 0 and 1. Update the position of each particle based on the calculated velocity of each particle. Use the formula x(t+1) = x(t) + v(t+1). Check whether the updated position meets the constraint condition, and correct it if it exceeds the boundary. Re-evaluate the fitness of the updated particle, and compare the fitness of the new position with the individual optimal position pi. If the new position is better, update pi. Compare pi of all particles, and update the global optimal position pg. Check the termination condition. It can be that the maximum number of iterations is reached, or the improvement amplitude of the global optimal solution is less than the preset threshold. If the termination condition is met, output the optimal solution pg; otherwise, return to the step of updating the velocity of each particle for iteration. Perform local search optimization on the optimal solution pg. Use the pattern search method to explore a small range around pg. Try in the positive and negative directions of each dimension, and update pg if a better solution is found. Repeat this process until no better solution can be found. Verify the feasibility of the final optimization result. Substitute the optimized control parameters into the system model, and perform simulation verification. Check whether all constraint conditions are met, and calculate the performance indicators. If the requirements are not met, adjust the optimization parameters and return to the step of obtaining the comprehensive performance evaluation result and updating the objective function of parameter optimization to re-execute the optimization process. Output the final optimization result, including the optimal control parameter value, the objective function value, and the performance indicators. These results will be used to update the control strategy and system configuration.

[0130] In this embodiment, by such a setting, by introducing system performance evaluation and self-optimization mechanism, including the calculation of multi-dimensional performance indicators, the application of fuzzy comprehensive evaluation, the update of optimization objectives, the implementation of improved particle swarm optimization algorithm and other innovative technologies, the continuous optimization and performance improvement of the power quality comprehensive management system are realized, the intelligent level and self-adaptive ability of the power quality comprehensive management system are enhanced, and strong technical support is provided for the long-term stable operation and continuous optimization of the power grid. First, the calculation of multi-dimensional performance indicators, such as voltage deviation, total harmonic distortion, voltage unbalance degree and dynamic response time, provides a comprehensive evaluation of system performance. This multi-angle evaluation method can capture various aspects of power quality problems and avoid the one-sidedness that may be caused by a single indicator. The application of fuzzy comprehensive evaluation method can handle the uncertainty and fuzziness in performance evaluation, and give more reasonable and comprehensive evaluation results. In the specific scene of power quality comprehensive management, firstly, they realize the comprehensive and objective evaluation of the performance of power quality control system. Secondly, the application of fuzzy comprehensive evaluation method greatly improves the reliability and practicality of the evaluation results. Especially in dealing with some difficult-to-quantify performance indicators, such as system stability margin or response ability to sudden events, fuzzy evaluation can give more realistic evaluation results. In addition, the self-optimization mechanism of this step also significantly improves the long-term reliability and efficiency of the system. Through continuous performance monitoring and parameter optimization, the system can continuously adapt to factors such as equipment aging and load characteristic changes, and maintain the best operating state. This not only improves the power quality, but also reduces the operating cost and maintenance demand of the system. Finally, this self-optimization mechanism provides the power quality management system with the ability to learn and evolve. Through the accumulation of long-term operation data and optimization experience, the system can continuously improve its control strategy and evaluation standard, and gradually improve its ability to handle complex power quality problems.

[0131] In further embodiments, further comprising: based on the performance evaluation results in the performance report, performing system adaptive optimization. Through online parameter identification, adaptive control law updating, multi-objective optimization, and robustness analysis, continuous optimization and performance improvement of the system are achieved. Online parameter identification is performed using pre-processed high-quality voltage data and control outputs of the target device control strategy. Recursive least squares method is used to update system model parameters in real time. First, an error equation for parameter estimation is constructed. Then, Kalman gain is calculated to update the parameter estimation value. The covariance matrix is updated to reflect the uncertainty of parameter estimation. Finally, the updated system model parameters, including the elements of the state transition matrix and the input matrix, are obtained. Based on the identified system parameters, the adaptive control law is updated. Model reference adaptive control method is used to design the reference model and define the tracking error. Lyapunov function is constructed to ensure system stability. The parameter adaptive law is derived to make the actual system output gradually close to the reference model output. The controller parameters, including feedback gain and feedforward gain, are updated. This adaptive mechanism can make the control system adapt to changes in system parameters and external disturbances. Using the updated objective function in the above performance evaluation process, multi-objective optimization is performed. Improved non-dominated sorting genetic algorithm (NSGA-III) is used to consider multiple objectives such as voltage stability, harmonic suppression, and energy efficiency. The population is initialized, and each individual represents a set of control parameters. Cross and mutation operations are performed to generate new offspring. According to the non-dominated sorting and crowding degree calculation, elite individuals are selected. Iterative evolutionary process is used to obtain a set of Pareto optimal solutions. This set of solutions reflects the trade-off between different objectives, providing multiple choices for decision-making. Robustness analysis is performed on the obtained Pareto optimal solutions. Monte Carlo simulation method is used to randomly sample system parameters and external disturbances. For each set of Pareto optimal solutions, simulation is performed under different parameter and disturbance conditions. The statistical characteristics of performance indicators, including mean, variance, and probability distribution, are calculated. The stability and performance degradation of each solution under parameter uncertainty and disturbance are evaluated. The solution with the best comprehensive performance and robustness is selected as the final control parameter.

[0132] Based on the robustness analysis results, update the control strategy. Introduce parameter uncertainty description to the prediction model of model predictive control to improve the accuracy of prediction. Add a robustness term to the objective function to balance the nominal performance and the worst-case performance. Update the constraint conditions to consider the impact of parameter variations. Modify the solving algorithm by using robust optimization techniques such as minimax method or probabilistic constraint method. These updates will improve the performance of the control system under uncertainty and disturbances. Apply the updated control strategy to the actual system. First, conduct comprehensive testing in a simulation environment to verify the effectiveness and safety of the new strategy. Then, design a smooth transition mechanism to gradually introduce the new control parameters and algorithms into the actual system. Monitor key performance indicators including voltage stability, harmonic content, and dynamic response characteristics. Record the system response and compare it with previous performance for analysis. If abnormalities occur, trigger a rollback mechanism immediately to restore to the previous stable configuration. Based on the actual operation data, perform long-term performance analysis and predictive maintenance. Use time series analysis methods such as autoregressive integrated moving average model (ARIMA) to analyze trends and predict key performance indicators. Apply machine learning algorithms such as support vector regression or random forest to establish a device health status evaluation model. Based on the analysis results, develop a predictive maintenance plan including equipment inspection, parameter tuning, and component replacement recommendations. This data-based maintenance strategy can improve the long-term reliability and efficiency of the system.

[0133] The technical scheme of the embodiment of the application is as follows: original three-phase voltage data of a transformer is acquired, pre-processing is performed on the original three-phase voltage data to obtain pre-processed three-phase voltage data; voltage disturbance features are extracted from the pre-processed three-phase voltage data; abnormality detection is performed based on the voltage disturbance features, a voltage disturbance classification result is determined through the abnormality detection result and the voltage disturbance features; and a target device control strategy is determined based on the voltage disturbance classification result, so as to control power quality through the target device control strategy. According to the technical scheme, the target device control strategy can be determined according to analysis of the voltage disturbance features, accurate control of power quality problems is achieved, and the accuracy of power quality treatment is improved.

[0134] Embodiment Three

[0135] Figure 3 is a structural schematic diagram of an electric energy quality comprehensive treatment system according to Embodiment Three of the application. The electric energy quality comprehensive treatment system can be used to execute the electric energy quality control method based on disturbance classification of any of the above embodiments. As shown in Figure 3As shown, this comprehensive power quality management system mainly includes a three-phase three-winding transformer, a back-to-back converter, and a bypass switch. The three windings of the three-winding transformer are the primary winding D1, the auxiliary circuit winding D2, and the main circuit winding Y1. The output of D2 serves as the AC input to the rectifier stage of the back-to-back converter, and each phase of Y1 is connected to the load via series connection with the inverter stage output of the back-to-back converter. The bypass switch is used to switch the inverter stage output.

[0136] In this embodiment, by controlling the back-to-back converter in the auxiliary circuit, power quality compensation and support functions not available in traditional transformers can be achieved. Based on the two-stage structure and functional characteristics of the back-to-back converter, the working principles of the rectifier stage and inverter stage can be analyzed separately. The rectifier stage is connected to D2 via a filter inductor and coupled in parallel to the load circuit through a transformer, providing reactive power compensation to the grid when necessary. Simultaneously, as the front stage of the back-to-back converter, the rectifier stage needs to provide stable DC voltage support to the inverter stage. In this invention, the inverter stage converter is a three-phase four-arm topology, with the fourth arm as the midpoint, and the three phases are independent. The output of the inverter stage is directly connected in series to the power supply circuit of the sensitive load, which can regulate the voltage in the main circuit, providing voltage support to the sensitive load and maintaining load voltage stability. The activation of inverter stage compensation can be controlled by controlling the opening and closing of the bypass switch.

[0137] The main control objectives of the rectifier-stage converter are twofold: first, to maintain the DC-side capacitor voltage at the reference value; and second, to absorb or generate reactive power to compensate the grid by controlling the output reactive current. To achieve these two control objectives, this paper adopts a dual-closed-loop decoupled PI control structure with an outer voltage loop and an inner current loop.

[0138] The design steps for the rectifier stage voltage outer loop control strategy are as follows:

[0139] Based on the low-frequency model of a three-phase two-level rectifier, when the switching frequency is much higher than the fundamental frequency of the power grid, the harmonic components in the PWM can be ignored. In this case, the switching function s k (k = a, b, c) can be expressed as

[0140]

[0141] In the formula, m is the modulation ratio of PWM, θ is the initial phase angle of the fundamental wave of the switching function, and ω is the angular frequency of the fundamental wave of the grid voltage.

[0142] For unity power factor sinusoidal current control, the grid-side current of the three-phase two-level rectifier is:

[0143]

[0144] In the formula, I m This indicates the peak value of the grid-side current.

[0145] In addition, the three-phase two-level rectifier DC side current i dc The following relationship exists between the switching function and the current:

[0146] i dc = s a i a + s b i b + s c i c ;

[0147] The DC side current can be approximately expressed as follows by combining the above equations:

[0148] i dc ≈ 0.75mcosθI m ;

[0149] 0.75mcosθ in the equation is a time-varying coefficient, and the maximum value 0.75 (m≤1) of the coefficient can be selected to replace the coefficient for the convenience of parameter design.

[0150] When a PI controller is applied to control the DC side voltage, the current loop transfer function can be approximated as a small inertia link, i.e.:

[0151]

[0152] In the equation, G V (s) represents the DC side voltage controller, G I (s) is the equivalent transfer function of the current loop closed loop, K v and T v are parameters of the voltage controller, and T s represents a sampling period.

[0153] At this time, the open loop transfer function of the voltage outer loop is:

[0154]

[0155] The voltage outer loop controller parameters can be designed according to the above equation.

[0156] The current inner loop control strategy design steps of the rectifier stage are as follows:

[0157] The dq model of the three-phase two-level rectifier is as follows:

[0158]

[0159] In the equation, e d and e q are dq axis components of the three-phase power grid voltage; v d and v qi represents the dq-axis components of the AC side voltage of a three-phase two-level rectifier; d and i q Let ρ be the dq-axis component of the AC side current of the three-phase two-level rectifier, and ρ be the differential operator.

[0160] As can be seen from the equation, coupling exists between the d and q axes, making controller design difficult. Therefore, this paper adopts a feedforward decoupling current control strategy, oriented with the d-axis, and the control equation is as follows:

[0161]

[0162] In the formula, KiP and KiI are the adjustment gain and integral adjustment gain of the current inner loop PI, respectively; i d * and i q * These are the reference values ​​for the grid-side current dq axis, respectively.

[0163] Considering the sampling delay and the inertial characteristics of PWM, after zero-pole correction according to the type I system, the open-loop transfer function of the inner current loop is:

[0164]

[0165] In the formula, KPWM is the equivalent gain of the bridge PWM of a three-phase two-level rectifier.

[0166] In this embodiment, the parameters of the current inner loop controller can be designed according to the above formula.

[0167] In this embodiment, the inverter stage of the back-to-back converter is responsible for providing voltage support when the load voltage drops, ensuring the stability of the sensitive load voltage. Therefore, the control strategy for the inverter stage is mainly divided into two parts: one is the grid connection point voltage detection algorithm, and the other is the inverter stage output control strategy. This paper selects voltage and current dual closed-loop control as the control method for the inverter stage output. In this embodiment, the grid connection point voltage detection process is as follows: the three-phase voltage at the grid connection point is subjected to Park transformation to obtain the three-phase voltage at the grid connection point in the dq coordinate system. After low-pass filtering, it is used to determine whether a voltage drop has occurred at the grid connection point: if no voltage drop has occurred, no compensation is performed; if a voltage drop has occurred, inverter stage compensation is initiated, the compensation voltage reference value in the dq coordinate system is calculated, and the inverter stage output is controlled to compensate for the voltage drop on the load through voltage and current dual closed-loop control. In this embodiment, the inverter stage output control strategy uses a PI controller for the outer voltage loop and proportional control for the inner current loop. After obtaining the output control compensation voltage of the inverter stage converter in the dq coordinate system, the Park inverse transform is performed to obtain the control compensation voltage of the inverter stage converter in the three-phase stationary coordinate system. SVPWM modulation is then performed to obtain the switching control signal of the inverter stage converter.

[0168] In this embodiment, when the power quality control method based on disturbance classification is implemented using the topology shown in the power quality comprehensive management system, the data processing flow is as follows:

[0169] Step S310: Acquire the raw three-phase voltage data. Through a series of data processing steps, obtain the preprocessed three-phase voltage data, time-frequency representation, instantaneous phasors, and sequence components. Specifically, voltage data is first collected from the primary winding D1, auxiliary circuit winding D2, and main circuit winding Y1 of the three-phase three-winding transformer. Then, adaptive filtering and denoising are performed on the raw data, followed by time-frequency analysis and adaptive threshold denoising. Subsequently, a high-quality voltage signal is reconstructed, and finally, the instantaneous phasors and sequence components are calculated.

[0170] Step S320: Utilize the preprocessed high-quality three-phase voltage data, time-frequency representation, instantaneous phasors, and sequence components to extract and classify voltage disturbance features. Simultaneously, collect capacitor voltage data from the DC side of the back-to-back converter as additional feature input. Anomaly detection is performed by calculating instantaneous amplitude and phase, applying Stockwell transform, extracting time-frequency features, applying local binary mode, constructing feature vectors, and finally using support vector data description.

[0171] Step S330: Based on the anomaly detection results of step S2, the detected voltage disturbance is accurately identified and quantitatively analyzed. Furthermore, current data is collected from the rectifier and inverter stages of the back-to-back converter for auxiliary analysis. Accurate identification and quantitative description of the disturbance are achieved through wavelet singularity detection, multi-resolution phase space reconstruction, dynamic invariant calculation, empirical mode decomposition, Hilbert-Huang transform, and fuzzy inference.

[0172] Step S340: Based on the disturbance analysis results of step S3, an adaptive control strategy is generated. The state information of the bypass switch is collected to determine whether to activate inverter stage compensation. Precise control of power quality is achieved by constructing a state-space model, designing a robust observer, applying model predictive control, dynamically adjusting control weights, solving the optimization problem, and generating PWM control signals. Finally, the generated control signals are output to the rectifier and inverter stages of the back-to-back converter.

[0173] Step S350: Based on the results of steps S310 to S340, the system performance is evaluated and self-optimization is achieved. Voltage and current data from the load side are collected to calculate actual power quality indicators. Continuous system improvement is achieved through performance indicator calculation, fuzzy comprehensive evaluation, updating optimization objectives, parameter optimization, and performance report generation. Finally, the optimized parameter settings are output to the back-to-back converter and control system.

[0174] In this embodiment, the three-phase winding transformer (including primary winding D1, auxiliary circuit winding D2, and main circuit winding Y1) provides the foundation for the entire system. This enables the system to convert power between different voltage levels and provides a flexible interface for power quality control. In the data processing flow of Embodiment 2, the high-speed sampling and signal processing in step S1 are based on the voltage data acquired using this transformer structure. The back-to-back converter, including the rectifier stage and inverter stage, is the core hardware for realizing power quality control. The model predictive control (MPC) and PWM control signal generation in step S4 are specifically designed for this back-to-back converter. The rectifier stage is used for reactive power compensation and DC voltage support, while the inverter stage is used for voltage support, which is directly related to the control strategy design. The structure of the auxiliary circuit (D2 winding connected to the rectifier stage of the back-to-back converter) and the main circuit (Y1 winding connected to the load through the inverter stage output) provides a means to implement the control strategy. The control strategy designed in step S340 is based on this structure, achieving reactive power compensation by controlling the rectifier stage and voltage support by controlling the inverter stage. The bypass switch is used to switch the inverter stage output and plays a crucial role in the control strategy. In the control strategy of step S340, the system needs to decide whether to activate inverter stage compensation based on the voltage disturbance, which is achieved by controlling the bypass switch.

[0175] The control objectives and strategies for the inverter stage include DC-side capacitor voltage control and reactive power compensation. This directly corresponds to the model predictive control strategy designed in step S340, particularly in handling reactive power compensation and DC voltage stabilization. The inverter stage control strategy, including grid-connected point voltage detection and dynamic voltage support, is related to the disturbance identification and control strategies designed in steps S330 and S340. This functionality is enhanced through more advanced algorithms such as wavelet singularity detection and fuzzy inference. The voltage and current dual closed-loop control structure is consistent with the control strategy framework designed in step S340.

[0176] In summary, based on the hardware topology of the power quality comprehensive management system in this embodiment, the disturbance classification-based power quality control method, through advanced signal processing, intelligent identification algorithms, and optimized control strategies, achieves a more efficient and intelligent power quality comprehensive management system. This solution fully utilizes the advantages of the existing hardware structure, while significantly improving the system's performance and adaptability through innovative software algorithms.

[0177] Example 4

[0178] Figure 4 This is a schematic diagram of a power quality control device based on disturbance classification according to Embodiment 4 of the present invention. Figure 4 As shown, the device includes:

[0179] The data acquisition module 410 is used to acquire the original three-phase voltage data of the transformer and perform preprocessing operations on the original three-phase voltage data to obtain preprocessed three-phase voltage data.

[0180] Feature extraction module 420 is used to extract voltage disturbance features from the preprocessed three-phase voltage data;

[0181] The voltage disturbance classification result determination module 430 is used to perform anomaly detection based on voltage disturbance characteristics, and determine the voltage disturbance classification result through the anomaly detection result and the voltage disturbance characteristics;

[0182] The control strategy determination module 440 is used to determine the target device control strategy based on the voltage disturbance classification results, so as to control the power quality through the target device control strategy.

[0183] Optionally, the preprocessed three-phase voltage data includes the instantaneous phasors and sequence components of the three-phase voltage;

[0184] The data acquisition module 410 is specifically used to acquire the raw three-phase voltage data, filter the raw three-phase voltage data using a set filter to obtain filtered three-phase voltage data, perform wavelet packet transform on the filtered three-phase voltage data to obtain transformed three-phase voltage data, and determine the instantaneous phasor and sequence component of the three-phase voltage based on the transformed three-phase voltage data.

[0185] Optionally, the feature extraction module 420 includes:

[0186] The data time-frequency analysis unit is used to perform time-frequency analysis on the preprocessed three-phase voltage data to obtain the time-frequency analysis results.

[0187] The feature determination unit is used to determine the time-frequency feature matrix and texture features based on the time-frequency analysis results.

[0188] The instantaneous voltage data determination unit is used to determine the instantaneous amplitude and phase of the three-phase voltage based on the instantaneous phasors of the three-phase voltage.

[0189] The feature fusion unit is used to fuse features based on the time-frequency feature matrix, texture features, instantaneous amplitude, and phase to obtain voltage disturbance features.

[0190] Optionally, the feature determination unit is specifically used to determine the time-domain energy, frequency-domain energy, and time-frequency energy distribution data based on the time-frequency analysis results; to obtain the time-frequency feature matrix by combining the time-domain energy, frequency-domain energy, and time-frequency energy distribution data; and to determine the texture features based on the time-frequency analysis results and the texture analysis algorithm.

[0191] Optionally, the voltage disturbance classification result determination module 430 is specifically used for:

[0192] Input the voltage disturbance characteristics into the set data description model for anomaly detection and obtain the anomaly detection results;

[0193] If the degree of abnormality in the abnormality detection result is greater than the set threshold, wavelet singularity detection is performed on the preprocessed three-phase voltage data to determine the start and end times of the voltage disturbance.

[0194] Construct a multi-resolution phase space based on the start and end times of the voltage disturbance;

[0195] The dynamic invariants of the voltage signal are determined by using a multi-resolution phase space, and voltage disturbance classification is performed based on the dynamic invariants to obtain the voltage disturbance classification results.

[0196] Optionally, the control strategy determination module 440 is specifically used for:

[0197] Determine the state-space model and the state observer;

[0198] The initial device control strategy is determined based on the state-space model and the state observer.

[0199] The initial device control strategy is adjusted based on the voltage disturbance classification results to obtain the target device control strategy.

[0200] Optionally, the device also includes a performance evaluation module, used to determine system performance indicators based on the control results of power quality control by the target device control strategy after controlling power quality through the target device control strategy; and to perform a comprehensive performance evaluation based on the system performance indicators to obtain a performance report.

[0201] The power quality control device based on disturbance classification provided in this embodiment of the invention can execute the power quality control method based on disturbance classification provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method.

[0202] Example 5

[0203] Figure 5 This is a schematic diagram of an electronic device according to Embodiment 5 of the present invention. The electronic device 10 is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (such as helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0204] likeFigure 5 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 can also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0205] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0206] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as a power quality control method based on disturbance classification.

[0207] In some embodiments, the power quality control method based on disturbance classification can be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the power quality control method based on disturbance classification described above can be performed. Alternatively, in other embodiments, processor 11 can be configured to perform the power quality control method based on disturbance classification by any other suitable means (e.g., by means of firmware).

[0208] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0209] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0210] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0211] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0212] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or computing systems that include middleware components (e.g., application servers), or computing systems that include frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0213] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0214] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0215] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A power quality control method based on disturbance classification, characterized in that, include: Obtain the raw three-phase voltage data of the transformer, and perform preprocessing operations on the raw three-phase voltage data to obtain preprocessed three-phase voltage data; Voltage disturbance features are obtained by feature extraction from the preprocessed three-phase voltage data; Anomaly detection is performed based on the voltage disturbance characteristics, and the voltage disturbance classification result is determined by the anomaly detection result and the voltage disturbance characteristics. Based on the voltage disturbance classification results, a target device control strategy is determined so as to control the power quality through the target device control strategy; The process includes anomaly detection based on the voltage disturbance characteristics, and determining the voltage disturbance classification result based on the anomaly detection result and the voltage disturbance characteristics, including: The voltage disturbance characteristics are input into a set data description model for anomaly detection to obtain anomaly detection results; If the degree of abnormality in the anomaly detection result is greater than a set threshold, then wavelet singularity detection is performed on the preprocessed three-phase voltage data to determine the start and end times of the voltage disturbance; wherein the start and end times of the voltage disturbance correspond to the singular points obtained by wavelet singularity detection. A multi-resolution phase space is constructed based on the start and end times of the voltage disturbance; wherein, the multi-resolution phase space refers to the space for representing and analyzing the phase and amplitude of the signal at different resolutions, and is used to capture the dynamic characteristics of the signal at different time scales; The dynamic invariants of the voltage signal are determined using the multi-resolution phase space, and voltage perturbation classification is performed based on these dynamic invariants to obtain voltage perturbation classification results. The dynamic invariants include the correlation dimension, the maximum Lyapunov exponent, and the approximate entropy. The correlation dimension is obtained by estimating the limit of the ratio of the logarithm of the correlation integral to the logarithm of the scale at different scales. The maximum Lyapunov exponent is obtained by tracking the separation rate of similar trajectories in the phase space. The approximate entropy is obtained by comparing the differences in pattern matching probabilities at different embedding dimensions.

2. The method according to claim 1, characterized in that, The preprocessed three-phase voltage data includes the instantaneous phasors and sequence components of the three-phase voltage. Obtain raw three-phase voltage data, and perform preprocessing operations on the raw three-phase voltage data to obtain the final three-phase voltage data, including: Acquire the raw three-phase voltage data, and then use a set filter to filter the raw three-phase voltage data to obtain filtered three-phase voltage data. The filtered three-phase voltage data is subjected to wavelet packet transform to obtain the transformed three-phase voltage data. The instantaneous phasors and sequence components of the three-phase voltage are determined based on the transformed three-phase voltage data.

3. The method according to claim 2, characterized in that, Voltage disturbance features are obtained by feature extraction from the preprocessed three-phase voltage data, including: Time-frequency analysis was performed on the preprocessed three-phase voltage data to obtain the time-frequency analysis results; The time-frequency feature matrix and texture features are determined based on the time-frequency analysis results; The instantaneous amplitude and phase of the three-phase voltage are determined based on the instantaneous phasors of the three-phase voltage. Voltage disturbance features are obtained by feature fusion based on the time-frequency feature matrix, texture features, instantaneous amplitude, and phase.

4. The method according to claim 3, characterized in that, Based on the time-frequency analysis results, the time-frequency feature matrix and texture features are determined, including: Based on the time-frequency analysis results, determine the time-domain energy, frequency-domain energy, and time-frequency energy distribution data; A time-frequency feature matrix is ​​obtained by combining the time-domain energy, the frequency-domain energy, and the time-frequency energy distribution data. Texture features are determined based on the time-frequency analysis results and texture analysis algorithm.

5. The method according to claim 1, characterized in that, The target device control strategy is determined based on the voltage disturbance detection results, including: Determine the state-space model and the state observer; The initial device control strategy is determined based on the state-space model and the state observer. The initial device control strategy is adjusted based on the voltage disturbance classification results to obtain the target device control strategy.

6. The method according to claim 1, characterized in that, After controlling the power quality through the target device control strategy, the method further includes: Based on the control results of the power quality control strategy of the target device, determine the system performance indicators; A comprehensive performance evaluation is conducted based on the system performance indicators to obtain a performance report.

7. A power quality control device based on disturbance classification, characterized in that, include: The data acquisition module is used to acquire the original three-phase voltage data of the transformer and perform preprocessing operations on the original three-phase voltage data to obtain preprocessed three-phase voltage data. The feature extraction module is used to extract voltage disturbance features from the preprocessed three-phase voltage data. The voltage disturbance classification result determination module is used to perform anomaly detection based on the voltage disturbance features, and determine the voltage disturbance classification result through the anomaly detection result and the voltage disturbance features; The control strategy determination module is used to determine the target device control strategy based on the voltage disturbance classification results, so as to control the power quality through the target device control strategy; Specifically, the voltage disturbance classification result determination module is used to input the voltage disturbance features into a set data description model for anomaly detection, and obtain anomaly detection results. If the anomaly degree in the anomaly detection results is greater than a set threshold, wavelet singularity detection is performed on the preprocessed three-phase voltage data to determine the start and end times of the voltage disturbance. The start and end times of the voltage disturbance correspond to the singular points obtained from the wavelet singularity detection. A multi-resolution phase space is constructed based on the start and end times of the voltage disturbance. The multi-resolution phase space refers to the phase and amplitude of the signal at different resolutions. The multi-resolution phase space is used to capture the dynamic characteristics of the signal at different time scales. The dynamic invariants of the voltage signal are determined using this multi-resolution phase space, and voltage perturbation classification is performed based on these dynamic invariants to obtain the voltage perturbation classification results. The dynamic invariants include the correlation dimension, the maximum Lyapunov exponent, and the approximate entropy. The correlation dimension is obtained by estimating the limit of the ratio of the logarithm of the correlation integral to the logarithm of the scale at different scales. The maximum Lyapunov exponent is obtained by tracking the separation rate of similar trajectories in the phase space. The approximate entropy is obtained by comparing the differences in pattern matching probabilities at different embedding dimensions.

8. The apparatus according to claim 7, characterized in that, The preprocessed three-phase voltage data includes the instantaneous phasors and sequence components of the three-phase voltage. The data acquisition module is specifically used to acquire the raw three-phase voltage data, filter the raw three-phase voltage data using a set filter to obtain filtered three-phase voltage data, perform wavelet packet transform processing on the filtered three-phase voltage data to obtain transformed three-phase voltage data, and determine the instantaneous phasor and sequence component of the three-phase voltage based on the transformed three-phase voltage data.

9. The apparatus according to claim 7, characterized in that, The feature extraction module includes: The data time-frequency analysis unit is used to perform time-frequency analysis on the preprocessed three-phase voltage data to obtain time-frequency analysis results. A feature determination unit is used to determine a time-frequency feature matrix and texture features based on the time-frequency analysis results; A voltage instantaneous data determination unit is used to determine the instantaneous amplitude and phase of the three-phase voltage based on the instantaneous phasors of the three-phase voltage; The feature fusion unit is used to perform feature fusion based on the time-frequency feature matrix, texture features, instantaneous amplitude, and phase to obtain voltage disturbance features.

10. The apparatus according to claim 9, characterized in that, The feature determination unit is specifically used to determine time-domain energy, frequency-domain energy, and time-frequency energy distribution data based on the time-frequency analysis results; to obtain a time-frequency feature matrix by combining the time-domain energy, the frequency-domain energy, and the time-frequency energy distribution data; and to determine texture features based on the time-frequency analysis results and the texture analysis algorithm.

11. The apparatus according to claim 7, characterized in that, The control strategy determination module is specifically used to determine the state space model and the state observer; determine the initial device control strategy based on the state space model and the state observer; and adjust the initial device control strategy based on the voltage disturbance classification results to obtain the target device control strategy.

12. The apparatus according to claim 7, characterized in that, Also includes: The performance evaluation module is used to determine system performance indicators based on the control results of the power quality control through the target device control strategy after controlling the power quality; and to perform a comprehensive performance evaluation based on the system performance indicators to obtain a performance report.

13. An electronic device, characterized in that, The electronic device includes: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the power quality control method based on disturbance classification as described in any one of claims 1-6.

14. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that cause a processor to execute the power quality control method based on disturbance classification as described in any one of claims 1-6.

Images