Voltage sag source identification method and device, electronic equipment and storage medium

Abstract

The invention relates to the technical field of voltage sag source identification, in particular to a voltage sag source identification method and device, electronic equipment and a storage medium, and the method comprises the steps: obtaining a three-phase voltage signal of a voltage sag signal according to a historical operation log of a power system; decomposing the three-phase voltage signal based on a CEEMD algorithm to obtain a plurality of decomposed modal components, and calculating an approximate entropy corresponding to each modal component; the approximate entropy corresponding to each modal component is input into a preset recognition model for classification recognition, and a classification recognition result is obtained.According to the method, Gaussian white noise which is in positive and negative pairs and equal in amplitude is added into the three-phase signals, the three-phase signals to be detected are decomposed more sufficiently, and the signal decomposition efficiency is effectively improved; the CSSMV cost sensitive support vector machine algorithm is optimized and trained through the IHPO hunter prey algorithm, the performance of classifying and identifying the voltage sag source is improved, and the identification accuracy is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of voltage sag source identification, and in particular to a voltage sag source identification method and device, electronic equipment and storage medium. Background Art

[0002] Voltage sag is a transient phenomenon in the power grid system, which is manifested as the effective value of the voltage suddenly drops to between 10% and 90% of the rated voltage, and then returns to normal within 0.5 to 30 cycles. Voltage sag has become the focus of attention in the field of power quality research due to its high frequency, wide range and serious damage.

[0003] At present, the identification of voltage sag sources is mainly divided into two stages: feature extraction and pattern recognition. Feature extraction refers to the use of signal processing technology to extract the time domain, frequency domain and waveform characteristics of the voltage sag signal. Common methods include Fourier transform, wavelet transform, and S transform. Pattern recognition refers to the design of a classifier and the use of relevant features to classify and identify the type of voltage sag source. Common algorithms include support vector machine SVM, neural network algorithm, etc.

[0004] However, existing feature extraction methods have problems such as mutual constraints between time domain and frequency domain, difficulty in decomposition, and high computational complexity. Existing pattern recognition methods also have problems such as requiring a large amount of data training, taking a long time, and being prone to overfitting leading to inaccuracy. Summary of the invention

[0005] The purpose of the present invention is to provide a method and device for identifying a voltage sag source, an electronic device and a storage medium. The present invention adopts the following technical means:

[0006] A method for identifying a voltage sag source, comprising:

[0007] According to the historical operation log of the power system, the three-phase voltage signal of the voltage sag signal is obtained;

[0008] Decomposing the three-phase voltage signal based on the CEEMD algorithm to obtain multiple modal components after decomposition, and calculating the approximate entropy corresponding to each of the modal components;

[0009] The approximate entropy corresponding to each of the modal components is input into a preset recognition model for classification and recognition to obtain a classification and recognition result; the preset recognition model is a hyperplane model constructed after optimizing and training the CSSMV cost-sensitive support vector machine algorithm based on the IHPO hunter-prey algorithm.

[0010] In some embodiments of the present application, obtaining the three-phase voltage signal of the voltage sag signal according to the historical operation log of the power system specifically includes:

[0011] The time-three-phase voltage data table of the power system historical operation log within the set time period is analyzed to determine the three-phase voltage data whose voltage value decays within the set time step and the corresponding time period, and the three-phase voltage data within the time period is determined as the three-phase voltage signal of the voltage sag signal.

[0012] In some embodiments of the present application, the three-phase voltage signal is decomposed based on the CEEMD algorithm to obtain the decomposed multiple modal components specifically including:

[0013] The three-phase voltage signal includes a plurality of three-phase voltage sub-signals, Gaussian white noise is added to one of the three-phase voltage sub-signals, and the added data is decomposed to obtain a plurality of first modal components;

[0014] Taking an average value of the plurality of first modal components to obtain a first eigenmodal component;

[0015] Performing residual processing on the first eigenmodal component to obtain a modal component corresponding to the three-phase voltage sub-signal;

[0016] Repeat the above process to obtain the modal components corresponding to each three-phase voltage sub-signal.

[0017] In some embodiments of the present application, the calculation of the approximate entropy corresponding to each modal component is specifically:

[0018] Based on the decomposed multiple modal components, a similarity measure is calculated;

[0019] Based on the similarity measure, multiple modal components are constructed into a multidimensional vector, and the approximate entropy of the multidimensional vector is calculated;

[0020] Based on the data correspondence between the multidimensional vector and each modal component, the approximate entropy of each modal component is determined.

[0021] In some embodiments of the present application, after optimizing and training the CSSMV cost-sensitive support vector machine algorithm based on the IHPO hunter-prey algorithm, constructing a hyperplane model specifically includes:

[0022] Based on the IHPO hunter-prey algorithm, a prey movement model is constructed. Based on the prey movement model, the CSSMV cost-sensitive support vector machine is optimized and trained to construct the IHPO-CSSMV hyperplane model.

[0023] In some embodiments of the present application, the inputting the approximate entropy corresponding to each modal component into a preset recognition model for classification and recognition to obtain the classification and recognition result specifically includes:

[0024] The approximate entropy corresponding to each modal component is input into the IHPO-CSSMV hyperplane model to obtain the positional relationship with the hyperplane, and the classification and recognition results are determined based on the positional relationship.

[0025] A voltage sag source identification device corresponds to a voltage sag source identification method, comprising:

[0026] A data acquisition module, used to acquire the three-phase voltage signal of the voltage sag signal according to the historical operation log of the power system;

[0027] A data processing module, used for decomposing the three-phase voltage signal based on the CEEMD algorithm to obtain a plurality of modal components after decomposition, and calculating the approximate entropy corresponding to each of the modal components;

[0028] A classification and recognition module, used for inputting the approximate entropy corresponding to each of the modal components into a preset recognition model for classification and recognition, and obtaining a classification and recognition result;

[0029] The model building module is used to optimize the training of the CSSMV cost-sensitive support vector machine algorithm based on the IHPO hunter-prey algorithm and build a hyperplane model.

[0030] A voltage sag source identification device is provided, which decomposes the three-phase voltage signal based on the CEEMD algorithm to obtain multiple modal components after decomposition. The data processing module specifically includes:

[0031] The first submodule is used to add Gaussian white noise to one of the three-phase voltage sub-signals and decompose the added data;

[0032] The second submodule is used to average the multiple first modal components to obtain a first eigenmodal component, and perform residual processing on the first eigenmodal component to obtain a modal component corresponding to the three-phase voltage sub-signal.

[0033] An electronic device, comprising:

[0034] processor;

[0035] A memory having executable codes stored thereon, which, when executed by the processor, causes the processor to execute any of the methods described above.

[0036] A computer-readable storage medium stores executable codes thereon. When the executable codes are executed by a processor of an electronic device, the processor is caused to execute any of the methods described above.

[0037] The beneficial effects of the present application are as follows: the present invention adopts the CEEMD algorithm to decompose the three-phase voltage signal, and by adding positive and negative paired Gaussian white noise with equal amplitude to the three signals, the three-phase signal to be detected is more fully decomposed, thereby effectively improving the signal decomposition efficiency; the CSSMV cost-sensitive support vector machine algorithm is optimized and trained through the IHPO hunter-prey algorithm, and the IHPO-CSSMV hyperplane model is used to classify and identify the voltage sag source, thereby improving the performance of classifying and identifying the voltage sag source and effectively improving the recognition accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 It is a flow chart of a method for identifying a voltage sag source in an embodiment of the present invention. DETAILED DESCRIPTION

[0039] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art. It should be noted that, in the absence of conflict, the embodiments of the present invention and the features described in the embodiments can be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0040] like Figure 1 As shown, a method for identifying a voltage sag source includes:

[0041] S1: Obtain the three-phase voltage signal of the voltage sag signal according to the historical operation log of the power system;

[0042] Specifically, voltage sag refers to a transient phenomenon in the power grid system, which is manifested as the effective value of the voltage suddenly drops to between 10% and 90% of the rated voltage, and then returns to normal within 0.5 to 30 cycles. The signal of the voltage sag phenomenon is called a voltage sag signal.

[0043] S2: Decomposing the three-phase voltage signal based on the CEEMD algorithm to obtain multiple modal components after decomposition, and calculating the approximate entropy corresponding to each of the modal components;

[0044] Specifically, the approximate entropy aims to evaluate the probability of new patterns appearing in the signal. The more complex the time series is, the larger its approximate entropy value is. On the contrary, the more regular the time series is, the smaller the corresponding approximate entropy value is.

[0045] S3: Input the approximate entropy corresponding to each of the modal components into a preset recognition model for classification and recognition to obtain a classification and recognition result; the preset recognition model is a hyperplane model constructed after optimizing the CSSMV cost-sensitive support vector machine algorithm based on the IHPO hunter-prey algorithm.

[0046] Furthermore, the principle of the IHPO algorithm is that when a hunter chases an individual of the prey away from the group and adjusts its own position according to the average position of the prey, the prey will also dynamically adjust its own position to avoid the predator.

[0047] Specifically, positive and negative Gaussian white noise with equal amplitudes are added to the three signals to more fully decompose the three-phase signal to be detected, effectively improving the signal decomposition efficiency; the CSSMV cost-sensitive support vector machine algorithm is optimized and trained through the IHPO hunter-prey algorithm to build a high-performance classification recognizer, which can improve the recognition accuracy.

[0048] In some embodiments of the present application, obtaining the three-phase voltage signal of the voltage sag signal according to the historical operation log of the power system specifically includes:

[0049] The time-three-phase voltage data table of the power system historical operation log within the set time period is analyzed to determine the three-phase voltage data whose voltage value decays within the set time step and the corresponding time period, and the three-phase voltage data within the time period is determined as the three-phase voltage signal of the voltage sag signal.

[0050] In some embodiments of the present application, the three-phase voltage signal is decomposed based on the CEEMD algorithm to obtain the decomposed multiple modal components specifically including:

[0051] The three-phase voltage signal includes a plurality of three-phase voltage sub-signals, Gaussian white noise is added to one of the three-phase voltage sub-signals, and the added data is decomposed to obtain a plurality of first modal components;

[0052] Taking an average value of the plurality of first modal components to obtain a first eigenmodal component;

[0053] Performing residual processing on the first eigenmodal component to obtain a modal component corresponding to the three-phase voltage sub-signal;

[0054] Repeat the above process to obtain the modal components corresponding to each three-phase voltage sub-signal.

[0055] The steps of CEEMD decomposing voltage sag data include:

[0056] Add Gaussian white noise to the voltage sag data y(t) to get the new signal y(t)+(-1) q εv j(q=1,2). Use EMD to decompose the data and get C 1 :

[0057]

[0058] where r j is the residual component;

[0059] The modal components obtained by decomposition are averaged to obtain the first eigenmodal component of CEEMD decomposition.

[0060] Remove the residual:

[0061] In the formula, the significance of removing the residual is to solve the problem of the transfer of white noise from high frequency to low frequency in CEEMD;

[0062] Use in r 1 Gaussian white noise is added to (t) to generate a new signal as a carrier, and EMD decomposition is performed to obtain the first-order modal component D 1 , based on which the second eigenmode component can be obtained:

[0063]

[0064] Remove the residual:

[0065] Repeat the above steps until no further decomposition can be performed, and the algorithm terminates. At this time, the number of eigenmode components obtained is K, and y(t) is decomposed into:

[0066] In some embodiments of the present application, the calculation of the approximate entropy corresponding to each modal component is specifically:

[0067] Based on the decomposed multiple modal components, a similarity measure is calculated;

[0068] Based on the similarity measure, multiple modal components are constructed into a multidimensional vector, and the approximate entropy of the multidimensional vector is calculated;

[0069] Based on the data correspondence between the multidimensional vector and each modal component, the approximate entropy of each modal component is determined.

[0070] Assume that the IMF component of each phase voltage sag signal decomposition is A i =[A i1 ,A i2 ,…,A ij ](i=1,2,3;j=1,..N), then the calculation method of approximate entropy is as follows:

[0071] Define algorithm-related parameters m, r, where m is an integer representing the length of the comparison vector, and r is a real number representing the measure of "similarity";

[0072] Reconstruct the m-dimensional vector {X(1),X(2),...,X(N-m+1)}, where X(i) = [A ik ,A i(k+1) ,...,A i(k+m-1) ];

[0073] For 1≤i≤N-m+1, count the number of vectors that satisfy the following conditions:

[0074]

[0075] where d[X,X * ]=max|A ij -A * ij |, A ij is the element of vector X, d represents the distance between vectors X(i) and X(j), which is determined by the maximum difference between the corresponding elements, j∈[1,N-m+1];

[0076] make Then the calculation formula of approximate entropy is ApEn=Φ m (r)-Φ m+1 (r).

[0077] Among them, m+1 represents the number of subsequences (vectors). The number of subsequences is determined, the distance between vectors is calculated, the ratio of similar vector pairs is counted, and the entropy value is calculated.

[0078] In some embodiments of the present application, after optimizing and training the CSSMV cost-sensitive support vector machine algorithm based on the IHPO hunter-prey algorithm, constructing a hyperplane model specifically includes:

[0079] Based on the IHPO hunter-prey algorithm, a prey movement model is constructed. Based on the prey movement model, the CSSMV cost-sensitive support vector machine is optimized and trained to construct the IHPO-CSSMV hyperplane model.

[0080] The specific steps are as follows:

[0081] S1: Initialize population individuals:

[0082] x i =rand(1,d)*(ub-lb)+lb;

[0083] Among them, x i is the position of the hunter or prey, lb is the lower bound of the problem variable, ub is the upper bound of the problem variable, and d is the dimension of the problem variable.

[0084] S2: During the HPO process, set the threshold to β and R5, where R5 is a random quantity in [0, 1]. When R5 < β, the current search agent is regarded as a hunter; otherwise, it is regarded as prey. The implementation of the HPO algorithm mainly relies on the following two mechanisms:

[0085] The expression of the hunter hunting mechanism is:

[0086] x(t + 1) = x(t) + 0.5[(2CZP pos - x(t)) + (2(1 - C)Zμ - x(t))];

[0087] where x(t) and x(t + 1) represent the current and next positions of the hunter respectively, P pos is the position of the prey being chased by the current hunter, usually taken as the individual farthest from the population average position, μ is the average of all individual positions in the current population, C is the balance parameter between exploration and exploitation, Z is the adaptive parameter of the algorithm, and the calculation formulas for μ, C, and Z are respectively:

[0088]

[0089] C = 1 - t * (0.98 / MaxIt);

[0090] Z = R2 * IDX + R3 * (~IDX), IDX = R1 > C;

[0091] where R1 and R3 are random vectors in [0, 1], P is the index value of R1 < C, R2 is a random number in [0, 1], IDX is the index value of vector R1 that satisfies the condition (P == 0), C is the balance parameter between exploration and exploitation, whose value decreases from 1 to 0.02 during the iteration process, t is the current iteration number, and MaxIt is the maximum number of iterations;

[0092] The expression of the prey movement mechanism is: x i (t + 1) = T pos + CZcos(2πR4) × (T pos - x i (t));

[0093] In the formula, T pos is the global optimal position, R4 is a random number in [-1, 1], and C and Z are the previous parameters;

[0094] However, the HPO algorithm is prone to falling into local optimal solutions. Therefore, the idea of dichotomy and recursion is introduced, that is, take:

[0095]

[0096] x = center + r * R;

[0097] newub = x, newlb = lb if R2 > 0

[0098] newub = ub, newlb = x else;

[0099] Where ub represents the upper bound, lb represents the lower bound, center represents the center point, r represents half of the distance between the upper bound and the lower bound, x represents a number within the interval, both R and R2 represent a random number with a value between -1 and 1, newub represents the new upper bound, newlb represents the new lower bound, and the original ub and lb are recorded during the recursive operation. Take newub and newlb as the new upper and lower bounds and pass them into the next recursive operation until newub - newlb < b, then reset newub and newlb to the original ub and lb, where b is the threshold. For example, take ub = 1, lb = -1, r equals 1, center is 0. When R randomly takes 0.5, x is 0.5 and is saved to the list. When R2 randomly takes -0.1, newub is 1 and newlb is 0.5. At this time, take b as 0.1, and the new interval length newub - newlb is 0.5, which is greater than 0.1. Repeat the above process until the new interval length is less than 0.1. Using this algorithm can initialize the population position.

[0100] In some embodiments of the present application, the inputting the approximate entropy corresponding to each modal component into a preset recognition model for classification recognition to obtain the classification recognition result specifically includes:

[0101] Input the approximate entropy corresponding to each modal component into the IHPO-CSSMV hyperplane model to obtain the positional relationship with the hyperplane, and determine the classification recognition result based on the positional relationship.

[0102] Specifically, the expression of the constructed hyperplane model is:

[0103]

[0104] In the formula, α i , α j represent Lagrange multipliers, y i , y j represent sample labels, x i , x j represent feature vectors, K(x i , x j ) represents the kernel function, l represents the number of samples, C + , C - represent penalty parameters.

[0105] A voltage sag source identification device, corresponding to a voltage sag source identification method, includes:

[0106] A data acquisition module, used to acquire the three-phase voltage signal of the voltage sag signal according to the historical operation log of the power system;

[0107] A data processing module, used for decomposing the three-phase voltage signal based on the CEEMD algorithm to obtain a plurality of modal components after decomposition, and calculating the approximate entropy corresponding to each of the modal components;

[0108] A classification and recognition module, used for inputting the approximate entropy corresponding to each of the modal components into a preset recognition model for classification and recognition, and obtaining a classification and recognition result;

[0109] The model building module is used to optimize the training of the CSSMV cost-sensitive support vector machine algorithm based on the IHPO hunter-prey algorithm and build a hyperplane model.

[0110] A voltage sag source identification device is provided, which decomposes the three-phase voltage signal based on the CEEMD algorithm to obtain multiple modal components after decomposition. The data processing module specifically includes:

[0111] The first submodule is used to add Gaussian white noise to one of the three-phase voltage sub-signals and decompose the added data;

[0112] The second submodule is used to average the multiple first modal components to obtain a first eigenmodal component, and perform residual processing on the first eigenmodal component to obtain a modal component corresponding to the three-phase voltage sub-signal.

[0113] An electronic device, comprising:

[0114] processor;

[0115] A memory having executable codes stored thereon, which, when executed by the processor, causes the processor to execute any of the methods described above.

[0116] A computer-readable storage medium stores executable codes thereon. When the executable codes are executed by a processor of an electronic device, the processor is caused to execute any of the methods described above.

[0117] The beneficial effects of the present application are as follows: the present invention adopts the CEEMD algorithm to decompose the three-phase voltage signal, and by adding positive and negative paired Gaussian white noise with equal amplitude to the three signals, the three-phase signal to be detected is more fully decomposed, thereby effectively improving the signal decomposition efficiency; the CSSMV cost-sensitive support vector machine algorithm is optimized and trained through the IHPO hunter-prey algorithm, and the IHPO-CSSMV hyperplane model is used to classify and identify the voltage sag source, thereby improving the performance of classifying and identifying the voltage sag source and effectively improving the recognition accuracy.

[0118] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for identifying a voltage sag source, characterized in that: include: According to the historical operation log of the power system, the three-phase voltage signal of the voltage sag signal is obtained; Decomposing the three-phase voltage signal based on the CEEMD algorithm to obtain multiple modal components after decomposition, and calculating the approximate entropy corresponding to each of the modal components; The approximate entropy corresponding to each of the modal components is input into a preset recognition model for classification and recognition to obtain a classification and recognition result; the preset recognition model is a hyperplane model constructed after optimizing and training the CSSMV cost-sensitive support vector machine algorithm based on the IHPO hunter-prey algorithm.

2. A method for identifying a voltage sag source according to claim 1, characterized in that: The method of obtaining the three-phase voltage signal of the voltage sag signal according to the historical operation log of the power system specifically includes: The time-three-phase voltage data table of the power system historical operation log within the set time period is analyzed to determine the three-phase voltage data whose voltage value decays within the set time step and the corresponding time period, and the three-phase voltage data within the time period is determined as the three-phase voltage signal of the voltage sag signal.

3. A method for identifying a voltage sag source according to claim 2, characterized in that: The three-phase voltage signal is decomposed based on the CEEMD algorithm to obtain the decomposed multiple modal components, which specifically include: The three-phase voltage signal includes a plurality of three-phase voltage sub-signals, Gaussian white noise is added to one of the three-phase voltage sub-signals, and the added data is decomposed to obtain a plurality of first modal components; Taking an average value of the plurality of first modal components to obtain a first eigenmodal component; Performing residual processing on the first eigenmodal component to obtain a modal component corresponding to the three-phase voltage sub-signal; Repeat the above process to obtain the modal components corresponding to each three-phase voltage sub-signal.

4. A method for identifying a voltage sag source according to claim 3, characterized in that: The calculation of the approximate entropy corresponding to each modal component is specifically as follows: Based on the decomposed multiple modal components, a similarity measure is calculated; Based on the similarity measure, multiple modal components are constructed into a multidimensional vector, and the approximate entropy of the multidimensional vector is calculated; Based on the data correspondence between the multidimensional vector and each modal component, the approximate entropy of each modal component is determined.

5. A method for identifying a voltage sag source according to claim 1, characterized in that: After the CSSMV cost-sensitive support vector machine algorithm is optimized and trained based on the IHPO hunter-prey algorithm, the hyperplane model is constructed specifically including: Based on the IHPO hunter-prey algorithm, a prey movement model is constructed. Based on the prey movement model, the CSSMV cost-sensitive support vector machine is optimized and trained to construct the IHPO-CSSMV hyperplane model.

6. A method for identifying a voltage sag source according to claim 5, characterized in that: The step of inputting the approximate entropy corresponding to each modal component into a preset recognition model for classification and recognition to obtain the classification and recognition result specifically includes: The approximate entropy corresponding to each modal component is input into the IHPO-CSSMV hyperplane model to obtain the positional relationship with the hyperplane, and the classification and recognition results are determined based on the positional relationship.

7. A voltage sag source identification device, corresponding to a voltage sag source identification method, characterized in that include: A data acquisition module, used to acquire three-phase voltage signals of voltage sag signals according to the historical operation log of the power system; A data processing module, used for decomposing the three-phase voltage signal based on the CEEMD algorithm to obtain a plurality of modal components after decomposition, and calculating the approximate entropy corresponding to each of the modal components; A classification and recognition module, used for inputting the approximate entropy corresponding to each of the modal components into a preset recognition model for classification and recognition, and obtaining a classification and recognition result; The model building module is used to optimize the training of the CSSMV cost-sensitive support vector machine algorithm based on the IHPO hunter-prey algorithm and build a hyperplane model.

8. A voltage sag source identification device, corresponding to decomposing the three-phase voltage signal based on the CEEMD algorithm to obtain multiple modal components after decomposition, characterized in that The data processing module specifically includes: The first submodule is used to add Gaussian white noise to one of the three-phase voltage sub-signals and decompose the added data; The second submodule is used to average the multiple first modal components to obtain a first eigenmodal component, and perform residual processing on the first eigenmodal component to obtain a modal component corresponding to the three-phase voltage sub-signal.

9. An electronic device, characterized in that: include: processor; A memory having executable codes stored thereon, which, when executed by the processor, causes the processor to execute the method according to any one of claims 1 to 6.

10. A computer-readable storage medium having executable codes stored thereon, which, when executed by a processor of an electronic device, causes the processor to execute the method according to any one of claims 1 to 6.

Images