A composite power quality disturbance signal identification method and device, electronic equipment and storage medium

Abstract

The present application relates to the technical field of composite power quality disturbance signal identification, and discloses a composite power quality disturbance signal identification method and device, electronic equipment and storage medium, the present application adopts synchronous compression S transform to process the disturbance signal, effectively compresses the time-frequency energy diffusion, improves the time-frequency aggregation degree of multiple superimposed disturbances, and the extracted statistical features have stronger discrimination; relying on the power quality disturbance identification model trained by the multiple signal-to-noise ratio and multiple composite disturbance sample, the labeling cost is reduced by fully utilizing the unlabeled data, and the stable identification accuracy is still maintained under the low noise working condition. The model supports online incremental updating, does not need full data retraining, the operation is accelerated by cholesky decomposition, can adapt to the real-time monitoring scene of the power grid, accurately distinguishes single, double and triple composite power disturbances, and improves the reliability and real-time performance of the power quality online identification.

Description

Technical Field

[0001] This invention relates to the field of composite power quality disturbance signal identification technology, specifically to a method, device, electronic device, and storage medium for identifying composite power quality disturbance signals. Background Technology

[0002] With the increasing penetration of renewable energy and power electronic loads, power systems face challenges in terms of power supply security and stability, and power quality disturbances have become more prominent than ever before. Typical power quality disturbance signals include voltage sags, voltage swells, harmonics, flicker, frequency deviations, and power outages. These issues stem from the intermittent nature of renewable energy generation and the nonlinear characteristics of modern loads, and are further exacerbated by the complex interactions between grid components. Therefore, power quality management has become an urgent task. The prerequisite for power quality control is the identification of power quality disturbances; thus, designing an effective power quality disturbance identification scheme is of great significance.

[0003] However, due to the resolution of time-frequency analysis of the signal and the lack of utilization of unlabeled data, the accuracy of power quality disturbance type identification is not high. Summary of the Invention

[0004] This invention provides a method, apparatus, electronic device, and storage medium for identifying composite power quality disturbance signals, in order to solve the problem that the accuracy of power quality disturbance type identification is not high due to the resolution of time-frequency analysis of signals and the lack of utilization of unlabeled data in the prior art.

[0005] In a first aspect, the present invention provides a method for identifying composite power quality disturbance signals, the method comprising: Acquire multiple types of composite power quality disturbance signals; The various composite power quality disturbance signals are subjected to synchronous compression S-transform to obtain the corresponding complex matrix; Using the feature vectors of each complex matrix, a training set is constructed and input into a preset initial power quality disturbance identification model for training, thereby generating a target power quality disturbance identification model; The real-time power quality disturbance signal of the power grid is collected and synchronously compressed by S-transform to obtain the corresponding real-time complex matrix; The eigenvector of the real-time complex matrix is ​​input into the target power quality disturbance identification model to identify the power quality disturbance type corresponding to the real-time power quality disturbance signal.

[0006] This invention employs synchronous compressed S-transform to process disturbance signals, effectively compressing time-frequency energy diffusion and improving the time-frequency aggregation of multiple superimposed disturbances, resulting in stronger discriminative power of extracted statistical features. It trains a power quality disturbance identification model based on samples containing multiple signal-to-noise ratios and various types of composite disturbances, fully utilizing unlabeled data to reduce labeling costs and maintaining stable identification accuracy even under low-noise conditions. The model supports online incremental updates without requiring full data retraining, and its computation is accelerated through Cholsky decomposition, making it adaptable to real-time power grid monitoring scenarios. It accurately identifies single, dual, and triple composite power disturbance types, improving the reliability and real-time performance of online power quality identification.

[0007] In one optional implementation, acquiring multiple types of composite power quality disturbance signals includes: Acquire multiple types of raw composite power quality disturbance signals; Gaussian white noise is superimposed on each of the original composite power quality disturbance signals to generate composite power quality disturbance signals with different signal-to-noise ratios.

[0008] This invention constructs a multi-noise intensity sample set by superimposing multiple Gaussian white noise levels (20dB, 30dB, 40dB, etc.) onto the original composite disturbance signal, simulating noise interference conditions in the power grid field. This enriches the diversity of training samples, enhances the model's resistance to noise interference, and avoids the model only adapting to ideal noise-free signals. This enables the algorithm to effectively extract the time-frequency features of disturbances even in harsh environments with low signal-to-noise ratios, ensuring stable and reliable identification results for various single and multiple composite disturbances.

[0009] In one optional implementation, the step of performing synchronous compression S-transform on each of the composite power quality disturbance signals to obtain the corresponding complex matrix includes: Perform an S-transform on each of the composite power quality disturbance signals to generate the corresponding initial complex matrix; Calculate the instantaneous frequency of the initial complex matrix; Based on the instantaneous frequency of the initial complex matrix, a synchronous compression operation is performed on the initial complex matrix to generate the final complex matrix.

[0010] This invention first obtains the complete time-frequency complex matrix of the signal through S-transform, and then solves the instantaneous frequency to complete the synchronous compression process, effectively converging time-frequency energy and suppressing spectral spread. Compared with the traditional S-transform, this method significantly improves the time-frequency clustering of single and multiple composite disturbances, weakens the aliasing of features of different disturbance components, facilitates the subsequent extraction of more identifiable statistical features, and enhances the distinction between complex superimposed disturbances and noisy signals.

[0011] In one optional implementation, the calculation process of the initial complex matrix is ​​as follows:

[0012] In the formula, This is a power quality disturbance signal. t It is time. It is the Gaussian window function of the signal. It is the frequency of the signal. The control parameters of the time axis determine the position of the Gaussian window on the time axis. It is a signal The initial complex matrix obtained after performing the S-transform.

[0013] The S-transform of this invention employs a frequency-adaptive Gaussian window function, which can adjust the window width according to the frequency, taking into account both time and frequency domain resolution. The output initial complex matrix completely preserves the time-frequency amplitude and phase information of the disturbance signal, providing a reliable raw time-frequency data foundation for subsequent instantaneous frequency calculation and synchronous compression operation.

[0014] In one optional implementation, the instantaneous frequency of the initial complex matrix is ​​calculated as follows:

[0015] In the formula, It is the initial complex matrix of the S-transform. It is to find the phase angle of a complex number. It involves finding the partial derivative with respect to time to obtain the instantaneous frequency of the complex matrix. It is the instantaneous frequency of the complex matrix.

[0016] This invention calculates the instantaneous frequency corresponding to each time-frequency point by performing phase calculation and time partial derivative operation on the initial complex matrix, obtains the true time-frequency distribution information of the signal, provides an accurate displacement reference for the frequency rearrangement of subsequent synchronous compression, and ensures the execution accuracy of time-frequency energy convergence compression.

[0017] In one optional implementation, the calculation process of the final complex matrix is ​​as follows:

[0018] In the formula, t It is the time axis of a complex matrix. It is the frequency axis of the new complex matrix. It is the Dirac function, i.e., the impulse function. It is the initial complex matrix obtained after the S-transformation. It is the new complex matrix obtained after synchronous compression of the complex matrix.

[0019] This invention utilizes the Dirac function to rearrange time-frequency energy, concentrating the diffused time-frequency energy to the corresponding position of the instantaneous frequency, suppressing spectral dispersion, obtaining a highly concentrated time-frequency complex matrix, enhancing the characteristic differences of various composite disturbances, and facilitating subsequent feature extraction and disturbance type identification.

[0020] In one optional implementation, the step of constructing a training set using the feature vectors of each of the complex matrices and inputting it into a preset initial power quality disturbance identification model for training, thereby generating a target power quality disturbance identification model, includes: Extract the multidimensional feature vectors of each complex matrix to construct a training set; Construct an initial power quality disturbance identification model; wherein, the objective function of the initial power quality disturbance identification model is:

[0021] In the formula, It is the output layer weight matrix. T These are the labels for the training set; H It is the output of the activation function; , These are the adjustment parameters of the objective function, used to make the objective function reach its minimum value; The training set is input into the initial power quality disturbance identification model for training to generate the target power quality disturbance identification model.

[0022] This invention introduces a dual regularization constraint term into the objective function of the power quality disturbance identification model. This suppresses overfitting of the weight matrix while fitting the training samples, and utilizes the trace operation term to mine the manifold structure of the samples, preserving the intrinsic neighborhood correlations of the samples. The trained target model has stronger generalization ability, reduces misjudgments caused by noisy samples, improves the accuracy of complex disturbance identification, and balances model computational efficiency, making it suitable for power grid disturbance identification tasks.

[0023] Secondly, the present invention provides a composite power quality disturbance signal identification device, the device comprising: The acquisition module is used to acquire multiple types of composite power quality disturbance signals; The transformation module is used to perform synchronous compression S-transformation on various composite power quality disturbance signals to obtain the corresponding complex matrix. The training module is used to construct a training set using the feature vectors of each complex matrix and input the preset initial power quality disturbance identification model for training, thereby generating the target power quality disturbance identification model. The acquisition module is used to acquire real-time power quality disturbance signals of the power grid and perform synchronous compressed S-transform to obtain the corresponding real-time complex matrix; The identification module is used to input the feature vector of the real-time complex matrix into the target power quality disturbance identification model to identify the power quality disturbance type corresponding to the real-time power quality disturbance signal.

[0024] Thirdly, the present invention provides an electronic device, comprising: a memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the computer instructions to perform the composite power quality disturbance signal identification method of the first aspect or any corresponding embodiment described above.

[0025] Fourthly, the present invention provides a computer-readable storage medium storing computer instructions for causing a computer to execute the composite power quality disturbance signal identification method of the first aspect or any corresponding embodiment described above. Attached Figure Description

[0026] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0027] Figure 1 This is a schematic flowchart of the first method for identifying composite power quality disturbance signals according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the second process of the composite power quality disturbance signal identification method according to an embodiment of the present invention; Figure 3 This is a schematic diagram of the synchronous compression S-transformation process according to an embodiment of the present invention; Figure 4 This is a schematic diagram of a power quality disturbance identification model according to an embodiment of the present invention; Figure 5 This is a structural block diagram of a composite power quality disturbance signal identification device according to an embodiment of the present invention; Figure 6 This is a schematic diagram of the hardware structure of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0029] It is understood that before using the technical solutions disclosed in the various embodiments of the present invention, users should be informed of the types, scope of use, and usage scenarios of the personal information involved in the present invention and their authorization should be obtained in accordance with relevant laws and regulations through appropriate means.

[0030] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0031] This invention provides a method for identifying composite power quality disturbance signals. By employing synchronous compressed S-transform processing of the disturbance signal, it effectively compresses time-frequency energy diffusion, improves the time-frequency aggregation of multiple superimposed disturbances, and enhances the discriminative power of the extracted statistical features. The power quality disturbance identification model is trained using samples with multiple signal-to-noise ratios and multiple types of composite disturbances, fully utilizing unlabeled data to reduce labeling costs and maintaining stable identification accuracy even under low-noise conditions. The model supports online incremental updates without requiring full data retraining. The computation is accelerated through Cholsky decomposition, making it adaptable to real-time power grid monitoring scenarios. It accurately identifies single, dual, and triple composite power disturbance types, improving the reliability and real-time performance of online power quality identification.

[0032] This embodiment provides a method for identifying composite power quality disturbance signals. Figure 1 This is a flowchart of a composite power quality disturbance signal identification method according to an embodiment of the present invention, such as... Figure 1 As shown, the process includes the following steps: Step S101: Obtain multiple types of composite power quality disturbance signals.

[0033] It should be noted that composite power quality disturbance signals refer to voltage / current signals that simultaneously contain two or more power quality disturbance components, covering double and triple superimposed disturbances, which are different from single disturbance signals such as voltage sags and harmonics, and are generated according to the IEEE-1159 standard.

[0034] In this embodiment of the invention, multiple types of composite power quality disturbance signals, including single disturbance, double superimposed disturbance, and triple superimposed disturbance, are obtained. The composite power quality disturbance signals contain simulated disturbance samples under different signal-to-noise ratios, providing the original signal data source for subsequent synchronous compression S-transform processing.

[0035] Step S102: Perform synchronous compression S-transform on various composite power quality disturbance signals to obtain the corresponding complex matrix.

[0036] It should be noted that synchronous compressed S-transform refers to a time-frequency analysis algorithm that, based on the standard S-transform, rearranges and compresses the energy of the time-frequency matrix by calculating the instantaneous frequency; the complex matrix refers to the two-dimensional time-frequency matrix output by synchronous compressed S-transform, where each element is a complex number and stores the amplitude and phase information of the signal at the corresponding time and frequency.

[0037] In this embodiment of the invention, the S-transform is sequentially performed on various composite power quality disturbance signals to solve the initial complex matrix, calculate the instantaneous frequency, and perform synchronous compression and rearrangement to obtain the corresponding complex matrix with improved time-frequency clustering, providing a time-frequency domain data foundation for subsequent feature vector extraction.

[0038] Step S103: Using the feature vectors of each complex matrix, a training set is constructed and input into a preset initial power quality disturbance identification model for training, thereby generating a target power quality disturbance identification model.

[0039] It should be noted that the feature vector refers to a one-dimensional vector formed by extracting statistical indicators such as maximum value, mean, variance, energy, skewness, and kurtosis from the magnitude of the synchronously compressed time-frequency complex matrix, which serves as the input data for the model; the training set refers to a sample set composed of feature vectors corresponding to multiple classes and multiple signal-to-noise ratio composite disturbance signals, along with the corresponding disturbance category labels; the initial power quality disturbance identification model refers to the basic online semi-supervised extreme learning machine model that has not undergone sample training and has only completed network structure and parameter initialization; and the target power quality disturbance identification model refers to a mature classification model that has undergone iterative optimization using the training set and whose output layer weights have converged, and can be directly used for real-time disturbance identification.

[0040] In this embodiment of the invention, multidimensional statistical feature vectors corresponding to each complex matrix are extracted, and corresponding disturbance category labels are matched for each feature vector to construct a training set. The training set is input into a preset initial power quality disturbance identification model to solve the objective function and complete the weight optimization training, thereby generating a converged target power quality disturbance identification model, which provides a usable model for subsequent real-time signal identification.

[0041] Step S104: Collect the real-time power quality disturbance signal of the power grid and perform synchronous compressed S-transform to obtain the corresponding real-time complex matrix.

[0042] It should be noted that real-time power quality disturbance signals refer to the real-time voltage and current waveforms collected online at the power grid site, including on-site noise and various superimposed disturbances, used for online detection and identification; real-time complex matrix refers to the time-frequency complex matrix obtained after performing synchronous compression S-transform on the real-time disturbance signals collected from the power grid, used to extract the features required for real-time identification.

[0043] In this embodiment of the invention, real-time power quality disturbance signals during the operation of the power grid are collected, and the same synchronous compression S-transform processing procedure is used to perform time-frequency analysis processing on the real-time signal to obtain the corresponding real-time complex matrix, which provides real-time time-frequency domain data for subsequent feature vector extraction and disturbance type identification.

[0044] Step S105: Input the feature vector of the real-time complex matrix into the target power quality disturbance identification model to identify the power quality disturbance type corresponding to the real-time power quality disturbance signal.

[0045] It should be noted that the power quality disturbance type refers to the signal category classified according to the form of disturbance superposition. This scheme has a total of 24 categories, including normal signals, single disturbances, double disturbances, and triple superposition disturbances.

[0046] In this embodiment of the invention, a feature vector with the same dimension as the training sample is extracted from the real-time complex matrix. The feature vector is then input into the trained target power quality disturbance identification model for classification operations. The classification result is output to identify the power quality disturbance type corresponding to the real-time power quality disturbance signal, thereby completing the online identification of power grid disturbances.

[0047] This embodiment provides a method for identifying composite power quality disturbance signals. Figure 2 This is a flowchart of a composite power quality disturbance signal identification method according to an embodiment of the present invention, such as... Figure 2 As shown, the process includes the following steps: Step S201: Obtain multiple types of composite power quality disturbance signals.

[0048] Specifically, step S201 includes: Step S2011: Obtain multiple types of original composite power quality disturbance signals.

[0049] In this embodiment of the invention, 24 sets of composite power quality disturbance signals are obtained. The signal parameters are uniformly distributed within the range specified by IEEE-1159, including 1 type of normal signal, 8 types of single disturbance power quality signals, 12 types of double disturbance power quality signals, and 3 types of triple disturbance power quality signals.

[0050] Step S2012: Gaussian white noise is superimposed on each original composite power quality disturbance signal to generate composite power quality disturbance signals with different signal-to-noise ratios.

[0051] In this embodiment of the invention, Gaussian white noise is added to 24 sets of composite power quality disturbance signals, with signal-to-noise ratios of 20dB, 30dB, and 40dB, respectively, to obtain multiple sets of composite power quality disturbance signals with different signal-to-noise ratios. Features are extracted from each set of power quality disturbance signals with different signal-to-noise ratios, and these features are used as a set of training data.

[0052] Step S202: Perform synchronous compression S-transform on various composite power quality disturbance signals to obtain the corresponding complex matrix.

[0053] Specifically, step S202 includes: Step S2021: Perform S-transform on each composite power quality disturbance signal to generate the corresponding initial complex matrix.

[0054] In specific implementations, such as Figure 3 As shown, the S-transform is performed on the training data of composite power quality disturbance signals with different signal-to-noise ratios to obtain the time-frequency characteristic complex matrix (i.e., the initial complex matrix) of the training data of composite power quality disturbance signals.

[0055] In some alternative implementations, the initial complex matrix is ​​calculated as follows:

[0056] In the formula, This is a power quality disturbance signal. t It is time. It is the Gaussian window function of the signal. It is the frequency of the signal. The control parameters of the time axis determine the position of the Gaussian window on the time axis. It is a signal The initial complex matrix obtained after performing the S-transform.

[0057] Step S2022: Calculate the instantaneous frequency of the initial complex matrix.

[0058] In this embodiment of the invention, the time-frequency characteristic complex matrix is ​​obtained by performing S-transform on composite power quality disturbance signals with different signal-to-noise ratios, and the instantaneous frequency of the time-frequency characteristic complex matrix is ​​further calculated.

[0059] In some optional implementations, the instantaneous frequency of the initial complex matrix is ​​calculated as follows:

[0060] In the formula, It is the initial complex matrix of the S-transform. It is to find the phase angle of a complex number. It involves finding the partial derivative with respect to time to obtain the instantaneous frequency of the complex matrix. It is the instantaneous frequency of the initial complex matrix.

[0061] Step S2023: Based on the instantaneous frequency of the initial complex matrix, perform synchronous compression operation on the initial complex matrix to generate the final complex matrix.

[0062] In this embodiment of the invention, after calculating the instantaneous frequency of the initial complex matrix, a synchronous compression operation is performed on the initial complex matrix to obtain the final complex matrix.

[0063] In some alternative implementations, the final complex matrix is ​​calculated as follows:

[0064] In the formula, t It is the time axis of a complex matrix. It is the frequency axis of the new complex matrix. It is the Dirac function, i.e., the impulse function. It is the initial complex matrix obtained after the S-transformation. It is the new complex matrix obtained after synchronous compression of the complex matrix.

[0065] Step S203: Using the feature vectors of each complex matrix, a training set is constructed and input into the preset initial power quality disturbance identification model for training, thereby generating the target power quality disturbance identification model.

[0066] Specifically, step S203 includes: Step S2031: Extract the multidimensional feature vectors of each complex matrix and construct the training set.

[0067] In this embodiment of the invention, the eigenvalues ​​of the complex matrix are extracted and used to construct the eigenvector:

[0068]

[0069]

[0070]

[0071]

[0072]

[0073]

[0074] In the formula, It is an element of a complex matrix, or a vector of a particular row within it. Max It is the maximum value of the modulus of the complex matrix. Min It is the minimum value of the modulus of the complex matrix. It is the average value of the magnitude of the complex matrix. It is the variance of the magnitude of the complex matrix. E It is the energy of the modulus of a complex matrix. Skewness It is the skewness value of the modulus of the complex matrix. Kurtosis It is the kurtosis value of the modulus of the complex matrix.

[0075] Training sample data is constructed using the multidimensional eigenvectors of the complex matrices of composite power quality disturbance signals with different signal-to-noise ratios.

[0076] Step S2032: Construct an initial power quality disturbance identification model; wherein, the objective function of the initial power quality disturbance identification model is:

[0077] In the formula, It is the output layer weight matrix. T These are the labels for the training set; H It is the output of the activation function; , These are the adjustment parameters of the objective function, used to minimize the objective function.

[0078] In embodiments of the present invention, such as Figure 4 As shown, the initial power quality disturbance identification model includes: an input layer for acquiring the feature vector of the extracted composite power quality disturbance signal; an input layer weight matrix for weighting the input feature vector; an activation function for simulating the response of neurons in a neural network to stimuli; an output layer weight matrix for weighting the output of the activation function; an output layer for outputting the category of the power quality disturbance signal; and an objective function, as shown in the formula:

[0079] In the formula, It is the output layer weight matrix. T These are the labels for the training set; H It is the output of the activation function; , These are the adjustment parameters of the objective function, used to minimize the objective function.

[0080] It is worth mentioning that the power quality disturbance identification model of the present invention adopts an online semi-supervised extreme learning machine model.

[0081] Step S2033: Input the training set into the initial power quality disturbance identification model for training, and generate the target power quality disturbance identification model.

[0082] In this embodiment of the invention, the initial online semi-supervised extreme learning machine model is trained using training sample data to obtain a pre-trained online semi-supervised extreme learning machine model, namely, the target power quality disturbance identification model.

[0083] Specifically, the training process of the online semi-supervised extreme learning machine algorithm is to minimize the objective function:

[0084] In the formula, For the objective function L Seeking information about The partial derivatives of the variables have the same meaning as in the previous section.

[0085] Setting the partial derivatives to zero, we can obtain the output layer weight matrix that minimizes the objective function. The formula is:

[0086] The above formula is the output layer weight matrix that minimizes the objective function.

[0087] Output layer weight matrix The iterative calculation formula is shown below:

[0088] In the formula, k represents the k-th iteration.

[0089] In a specific embodiment, Satisfy the following formula:

[0090] In the formula, It is the upper triangular matrix of the Chollisky decomposition of the matrix. It is a lower triangular matrix of the Chollisky decomposition of a matrix.

[0091] Step S204: Acquire the real-time power quality disturbance signal of the power grid and perform synchronous compressed S-transform to obtain the corresponding real-time complex matrix. For details, please refer to [link to relevant documentation]. Figure 1 Step S104 of the illustrated embodiment will not be described again here.

[0092] Step S205 involves inputting the eigenvectors of the real-time complex matrix into the target power quality disturbance identification model to identify the power quality disturbance type corresponding to the real-time power quality disturbance signal. For details, please refer to [link to relevant documentation]. Figure 1 Step S105 of the illustrated embodiment will not be described again here.

[0093] This invention extracts the feature vector of a composite power quality disturbance signal by synchronously compressing the S-transform. The feature vector is then used as input to an online semi-supervised extreme learning machine (OLSM) to obtain a pre-trained OLSM. The pre-trained model includes the following parameters: the hidden layer weight matrix, the number of activation functions, the output layer weight matrix, and the optimal control parameters of the objective function. This invention enables rapid solution of the time-frequency feature matrix of composite power quality disturbance signals, exploring a new technical method for the accurate identification of such signals.

[0094] This embodiment also provides a composite power quality disturbance signal identification device, which is used to implement the above embodiments and preferred embodiments; details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that implements a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0095] This embodiment provides a composite power quality disturbance signal identification device, such as... Figure 5 As shown, it includes: Acquisition module 301 is used to acquire multiple types of composite power quality disturbance signals; Transformation module 302 is used to perform synchronous compression S-transformation on various composite power quality disturbance signals to obtain the corresponding complex matrix; Training module 303 is used to construct a training set by using the feature vectors of each complex matrix and inputting it into a preset initial power quality disturbance identification model for training, thereby generating a target power quality disturbance identification model. The acquisition module 304 is used to acquire real-time power quality disturbance signals of the power grid and perform synchronous compressed S-transform to obtain the corresponding real-time complex matrix; The identification module 305 is used to input the feature vector of the real-time complex matrix into the target power quality disturbance identification model to identify the power quality disturbance type corresponding to the real-time power quality disturbance signal.

[0096] In some optional implementations, the acquisition module 301 includes: The acquisition unit is used to acquire multiple types of raw composite power quality disturbance signals; The superposition unit is used to superimpose Gaussian white noise onto each original composite power quality disturbance signal to generate composite power quality disturbance signals with different signal-to-noise ratios.

[0097] In some alternative implementations, the transformation module 302 includes: The S-transform unit is used to perform S-transform on each composite power quality disturbance signal to generate the corresponding initial complex matrix. The computational unit is used to calculate the instantaneous frequency of the initial complex matrix; The arithmetic unit is used to perform synchronous compression operations on the initial complex matrix based on the instantaneous frequency of the initial complex matrix to generate the final complex matrix.

[0098] In some alternative implementations, the initial complex matrix is ​​calculated as follows:

[0099] In the formula, This is a power quality disturbance signal. t It is time. It is the Gaussian window function of the signal. It is the frequency of the signal. The control parameters of the time axis determine the position of the Gaussian window on the time axis. It is a signal The initial complex matrix obtained after performing the S-transform.

[0100] In some optional implementations, the instantaneous frequency of the initial complex matrix is ​​calculated as follows:

[0101] In the formula, It is the initial complex matrix of the S-transform. It is to find the phase angle of a complex number. It involves finding the partial derivative with respect to time to obtain the instantaneous frequency of the complex matrix. It is the instantaneous frequency of the complex matrix.

[0102] In some alternative implementations, the final complex matrix is ​​calculated as follows:

[0103] In the formula, t It is the time axis of a complex matrix. It is the frequency axis of the new complex matrix. It is the Dirac function, i.e., the impulse function. It is the initial complex matrix obtained after the S-transformation. It is the new complex matrix obtained after synchronous compression of the complex matrix.

[0104] In some alternative implementations, training module 303 includes: The extraction unit is used to extract the multidimensional feature vectors of each complex matrix and construct the training set. The building unit is used to construct the initial power quality disturbance identification model; wherein, the objective function of the initial power quality disturbance identification model is:

[0105] In the formula, It is the output layer weight matrix. T These are the labels for the training set; H It is the output of the activation function; , These are the adjustment parameters of the objective function, used to make the objective function reach its minimum value; The training unit is used to input the training set into the initial power quality disturbance identification model for training, and generate the target power quality disturbance identification model.

[0106] The composite power quality disturbance signal identification device provided in this embodiment of the invention can execute the composite power quality disturbance signal identification method provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method. Further functional descriptions of the above modules and units are the same as those in the corresponding embodiments described above, and will not be repeated here.

[0107] Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention.

[0108] The following is a detailed reference. Figure 6 This diagram illustrates a structural schematic suitable for implementing an electronic device according to embodiments of the present invention. The electronic device may include a processor (e.g., a central processing unit, graphics processor, etc.) 401, which can perform various appropriate actions and processes according to a program stored in read-only memory (ROM) 402 or a program loaded from memory 408 into random access memory (RAM) 403. The RAM 403 also stores various programs and data required for the operation of the electronic device. The processor 401, ROM 402, and RAM 403 are interconnected via a bus 404. An input / output (I / O) interface 405 is also connected to the bus 404.

[0109] Typically, the following devices can be connected to I / O interface 405: input devices 406 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 407 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; memory devices 408 including, for example, magnetic tapes, hard disks, etc.; and communication devices 409. Communication device 409 allows electronic devices to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 6 Electronic devices with various devices are shown, but it should be understood that it is not required to implement or have all of the devices shown, and more or fewer devices may be implemented or have instead.

[0110] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device 409, or installed from a memory 408, or installed from a ROM 402. When the computer program is executed by the processor 401, it performs the functions defined in the composite power quality disturbance signal identification method of the embodiments of the present invention.

[0111] Figure 6The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments of the present invention.

[0112] This invention also provides a computer-readable storage medium. The methods described above according to embodiments of the invention can be implemented in hardware or firmware, or implemented as computer code that can be recorded on a storage medium, or implemented as computer code downloaded via a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and then stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium can also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code. When the software or computer code is accessed and executed by the computer, processor, or hardware, the composite power quality disturbance signal identification method shown in the above embodiments is implemented.

[0113] Although embodiments of the invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the invention, and all such modifications and variations fall within the scope defined by the appended claims.

Claims

1. A method for identifying composite power quality disturbance signals, characterized in that, The method includes: Acquire multiple types of composite power quality disturbance signals; The various composite power quality disturbance signals are subjected to synchronous compression S-transform to obtain the corresponding complex matrix; Using the feature vectors of each complex matrix, a training set is constructed and input into a preset initial power quality disturbance identification model for training, thereby generating a target power quality disturbance identification model; The real-time power quality disturbance signal of the power grid is collected and synchronously compressed by S-transform to obtain the corresponding real-time complex matrix; The eigenvector of the real-time complex matrix is ​​input into the target power quality disturbance identification model to identify the power quality disturbance type corresponding to the real-time power quality disturbance signal.

2. The method according to claim 1, characterized in that, The acquisition of multiple types of composite power quality disturbance signals includes: Acquire multiple types of raw composite power quality disturbance signals; Gaussian white noise is superimposed on each of the original composite power quality disturbance signals to generate composite power quality disturbance signals with different signal-to-noise ratios.

3. The method according to claim 1, characterized in that, The step of performing synchronous compression S-transform on each of the composite power quality disturbance signals to obtain the corresponding complex matrix includes: Perform an S-transform on each of the composite power quality disturbance signals to generate the corresponding initial complex matrix; Calculate the instantaneous frequency of the initial complex matrix; Based on the instantaneous frequency of the initial complex matrix, a synchronous compression operation is performed on the initial complex matrix to generate the final complex matrix.

4. The method according to claim 3, characterized in that, The calculation process of the initial complex matrix is ​​as follows: In the formula, This is a power quality disturbance signal. t It is time. It is the Gaussian window function of the signal. It is the frequency of the signal. The control parameters of the time axis determine the position of the Gaussian window on the time axis. It is a signal The initial complex matrix obtained after performing the S-transform.

5. The method according to claim 3, characterized in that, The calculation process for the instantaneous frequency of the initial complex matrix is ​​as follows: In the formula, It is the initial complex matrix of the S-transform. It is to find the phase angle of a complex number. It involves finding the partial derivative with respect to time to obtain the instantaneous frequency of the complex matrix. It is the instantaneous frequency of the complex matrix.

6. The method according to claim 3, characterized in that, The calculation process for the final complex matrix is ​​as follows: In the formula, t It is the time axis of a complex matrix. It is the frequency axis of the new complex matrix. It is the Dirac function, i.e., the impulse function. It is the initial complex matrix obtained after the S-transformation. It is the new complex matrix obtained after synchronous compression of the complex matrix.

7. The method according to claim 1, characterized in that, The step of constructing a training set using the feature vectors of each complex matrix and inputting it into a preset initial power quality disturbance identification model for training, thereby generating a target power quality disturbance identification model, includes: Extract the multidimensional feature vectors of each complex matrix to construct a training set; Construct an initial power quality disturbance identification model; wherein, the objective function of the initial power quality disturbance identification model is: In the formula, It is the output layer weight matrix. T These are the labels for the training set; H It is the output of the activation function; , These are the adjustment parameters of the objective function, used to make the objective function reach its minimum value; The training set is input into the initial power quality disturbance identification model for training to generate the target power quality disturbance identification model.

8. A composite power quality disturbance signal identification device, characterized in that, The device includes: The acquisition module is used to acquire multiple types of composite power quality disturbance signals; The transformation module is used to perform synchronous compression S-transformation on various composite power quality disturbance signals to obtain the corresponding complex matrix. The training module is used to construct a training set using the feature vectors of each complex matrix and input the preset initial power quality disturbance identification model for training, thereby generating the target power quality disturbance identification model. The acquisition module is used to acquire real-time power quality disturbance signals of the power grid and perform synchronous compressed S-transform to obtain the corresponding real-time complex matrix; The identification module is used to input the feature vector of the real-time complex matrix into the target power quality disturbance identification model to identify the power quality disturbance type corresponding to the real-time power quality disturbance signal.

9. An electronic device, characterized in that, include: A memory and a processor are communicatively connected, the memory stores computer instructions, and the processor executes the computer instructions to perform the composite power quality disturbance signal identification method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing the computer to execute the composite power quality disturbance signal identification method according to any one of claims 1 to 7.

Images