Pole-mounted circuit breaker data processing method, equipment, medium and product

Abstract

The invention discloses a pole-mounted circuit breaker data processing method and device, a medium and a product, and relates to the technical field of signal control and electrics, and the method comprises the steps: carrying out the decomposition and reconstruction of original vibration signal data through employing a modal decomposition algorithm, obtaining a plurality of reconstruction components, and enlarging the signal feature range through the decomposition and reconstruction of the data; performing data feature hierarchical clustering on the plurality of reconstruction components to obtain a reconstruction aggregation signal, and capturing a hierarchical relationship among the reconstruction components through hierarchical clustering; performing feature extraction on the reconstructed aggregated signal by using a first variational mode decomposition algorithm to obtain a high-feature reconstructed signal; the first variational mode decomposition algorithm is a variational mode decomposition algorithm after parameter optimization by using a zebra algorithm, and after the variational mode decomposition algorithm is optimized by using the zebra algorithm, feature extraction is performed on the reconstructed aggregated signal, so that data features are further improved. According to the invention, deep information of small sample signal data can be extracted.

Description

Technical Field

[0001] The present application relates to the fields of signal control and electrical technology, and in particular to a method, device, medium and product for processing data of a pole-mounted circuit breaker. Background Art

[0002] In today's data-driven era, small sample signal processing technology has a vital impact on economic and scientific research development. In key areas such as the various motion states of pole-mounted circuit breakers in power environments, obtaining large amounts of data is expensive and in some cases almost impossible, which makes the effective use of small sample data particularly critical. Through appropriate signal processing technology, even from a limited data set, valuable information can be extracted, providing a more accurate basis for subsequent research on pole-mounted circuit breakers, thereby reducing the R&D cost of the project and promoting sustainable economic development.

[0003] Traditional signal processing techniques, such as Fourier transform, have certain advantages in frequency domain analysis, but they are often difficult to effectively capture the local features of signals when processing non-stationary signals. In addition, when processing small sample signals with complex structures, existing feature extraction methods may not be able to fully explore the deep information in the signal, thus affecting subsequent research such as the judgment of data feature types. Summary of the invention

[0004] The purpose of this application is to provide a pole-mounted circuit breaker data processing method, device, medium and product, which can extract deep information from small sample signal data.

[0005] To achieve the above objectives, this application provides the following solutions:

[0006] In a first aspect, the present application provides a method for processing data of a pole-mounted circuit breaker, comprising:

[0007] Obtaining raw vibration signal data of the pole-mounted circuit breaker;

[0008] Decomposing and reconstructing the original vibration signal data using a modal decomposition algorithm to obtain a plurality of reconstructed components;

[0009] Performing data feature hierarchical clustering on the plurality of reconstruction components to obtain a reconstructed aggregate signal;

[0010] A first variational mode decomposition algorithm is used to extract features of the reconstructed aggregate signal to obtain a high-feature reconstructed signal; the first variational mode decomposition algorithm is a variational mode decomposition algorithm after parameter optimization using a zebra algorithm.

[0011] Optionally, the process of optimizing the variational mode decomposition algorithm parameters using the zebra algorithm includes:

[0012] With the goal of minimizing the mean value of sample entropy, the fitness function is used to evaluate each zebra individual in the zebra algorithm to obtain the optimal zebra position and the optimal fitness index;

[0013] The variational mode decomposition algorithm parameters are optimized according to the optimal zebra position and the optimal fitness index.

[0014] Optionally, performing data feature hierarchical clustering on the plurality of reconstruction components to obtain a reconstructed aggregate signal specifically includes:

[0015] Consider each of the reconstructed components as an initial cluster;

[0016] Calculate the inter-cluster distance between every two initial clusters;

[0017] hierarchical clustering is performed on the reconstructed components based on the maximum inter-cluster distance; the maximum inter-cluster distance is the maximum value of all the inter-cluster distances;

[0018] According to the inter-cluster distance and the inter-cluster maximum distance, the clusters with the closest distance are merged to obtain a reconstructed aggregate signal.

[0019] Optionally, the calculating the inter-cluster distance between every two initial clusters specifically includes:

[0020] The Chebyshev distance between each two initial clusters is calculated to obtain the inter-cluster distance.

[0021] Optionally, after extracting features from the reconstructed aggregate signal using the first variational mode decomposition algorithm to obtain a high-feature reconstructed signal, the method further includes:

[0022] Perform Fourier transform on the high-feature reconstructed signal to obtain spectrum data.

[0023] Optionally, performing data feature hierarchical clustering on the plurality of reconstruction components to obtain a reconstructed aggregate signal specifically includes:

[0024] The data features of the plurality of reconstruction components are hierarchically clustered using an aggregation hierarchical clustering algorithm to obtain a reconstructed aggregation signal.

[0025] Optionally, the objective function of the zebra algorithm is:

[0026]

[0027]

[0028] Among them, E i For each analytic signal, the sample entropy, H i To reconstruct the analytical signal obtained by Hilbert transform of the aggregate signal, m is the number of continuous data points considered when calculating the sample entropy, and j is the sample entropy position information in the sequence.

[0029] In a second aspect, the present application provides a computer device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement any of the above-mentioned pole-mounted circuit breaker data processing methods.

[0030] In a third aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the pole-mounted circuit breaker data processing method described in any one of the above is implemented.

[0031] In a fourth aspect, the present application provides a computer program product, including a computer program, which, when executed by a processor, implements any of the above-mentioned pole-mounted circuit breaker data processing methods.

[0032] According to the specific embodiments provided in this application, this application discloses the following technical effects:

[0033] The present application provides a method, device, medium and product for processing data of a pole-mounted circuit breaker. The original vibration signal data of the pole-mounted circuit breaker is decomposed and reconstructed by a modal decomposition algorithm to obtain several reconstructed components. The signal feature range of the small sample data is expanded by decomposing and reconstructing the data; the data features of the several reconstructed components are hierarchically clustered to obtain a reconstructed aggregate signal. The hierarchical relationship between the reconstructed components is captured by hierarchical clustering to generate a reconstructed aggregate signal with the same data dimension and different hierarchical structures; the reconstructed aggregate signal is feature extracted by a first variational modal decomposition algorithm to obtain a high-feature reconstructed signal; the first variational modal decomposition algorithm is a variational modal decomposition algorithm after parameter optimization by a zebra algorithm, and the variational modal decomposition algorithm is optimized by a zebra algorithm to extract features from the reconstructed aggregate signal, thereby further improving the data features. The present application can extract deep information of small sample signal data. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0035] Figure 1 This is an application environment diagram of a pole-mounted circuit breaker data processing method in one embodiment of the present application;

[0036] Figure 2 A schematic flow chart of a method for processing data of a pole-mounted circuit breaker provided in one embodiment of the present application;

[0037] Figure 3 A schematic diagram of a data feature extraction process based on AGCEEMDAN-ZOA-VMD provided in one embodiment of the present application;

[0038] Figure 4 A schematic diagram of a data dimension conversion process provided in an embodiment of the present application;

[0039] Figure 5 A schematic diagram of the structure of a computer device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0040] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0041] The present application proposes a data feature extraction method for a pole-mounted circuit breaker based on secondary decomposition of the data's lateral dimension. Based on the problem that data feature offset is caused by too small a data sample and subsequent classification and other tasks cannot be correctly identified, the vibration signal is firstly reconstructed in the time domain using a Completely Adaptive Noise Set Empirical Mode Decomposition (CEEMDAN) algorithm and an Agglomerative Hierarchical Clustering (AGG) algorithm to expand the range of its signal features. Based on the sample entropy of the reconstructed data, the reconstructed signal components (IMFs) are hierarchically clustered to capture the hierarchical relationship between the IMFs and generate aggregated signal components (co-IMFs) with the same data dimension and different hierarchical structures. Subsequently, the zebra optimization algorithm (ZOA) was combined with the variational mode decomposition algorithm (VMD), and a VMD modal fitness measurement index was proposed. Each co-IMF component was used as input to further improve the data characteristics. Finally, Fourier transformation was performed to expand the frequency domain characteristics of the data, integrating data acquisition, time-frequency domain feature expansion, and enhancement functions, so that subsequent operators can conduct efficient pole-mounted circuit breaker signal recognition, early warning, and other research. Therefore, this application proposes an efficient small sample feature processing technology to facilitate the judgment and research operations of pole-mounted circuit breaker data by workers and researchers on the basis of expanding the data feature range and feature quantity.

[0042] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0043] The pole-mounted circuit breaker data processing method provided in the embodiment of the present application can be applied to Figure 1 In the application environment shown. Among them, the terminal 102 communicates with the server 104 through the network. The data storage system can store the data that the server 104 needs to process. The data storage system can be set up separately, integrated on the server 104, or placed on the cloud or other servers. The terminal 102 can send the original vibration signal data of the pole-mounted circuit breaker to be processed to the server 104. After the server 104 receives the data to be processed, the server 104 uses a modal decomposition algorithm to decompose and reconstruct the original vibration signal data of the pole-mounted circuit breaker to obtain a number of reconstructed components; the data features of the several reconstructed components are hierarchically clustered to obtain a reconstructed aggregate signal; the reconstructed aggregate signal is extracted using a first variational modal decomposition algorithm to obtain a high-feature reconstructed signal; the first variational modal decomposition algorithm is a variational modal decomposition algorithm after parameter optimization using a zebra algorithm. The server 104 can feed back the obtained high-feature reconstructed signal to the terminal 102. In addition, in some embodiments, the pole-mounted circuit breaker data processing method can also be implemented independently by the server 104 or the terminal 102. For example, the terminal 102 can directly process the small sample data to be processed, or the server 104 can obtain the small sample data to be processed from the data storage system and process it.

[0044] The terminal 102 may be, but is not limited to, various desktop computers, laptop computers, smart phones, tablet computers, IoT devices, and portable wearable devices. The IoT devices may be smart speakers, smart TVs, smart air conditioners, smart vehicle-mounted devices, etc. The portable wearable devices may be smart watches, smart bracelets, head-mounted devices, etc. The server 104 may be implemented as an independent server or a server cluster consisting of multiple servers, or may be a cloud server.

[0045] In an exemplary embodiment, Figure 2 As shown, a method for processing data of a pole-mounted circuit breaker is provided. The method is executed by a computer device, and can be executed by a computer device such as a terminal or a server alone, or by a terminal and a server together. In the embodiment of the present application, the method is applied to Figure 1 The server 104 in the example is used as an example to illustrate the method, which includes the following steps S1 to S4. Among them:

[0046] S1. Obtaining raw vibration signal data of a pole-mounted circuit breaker.

[0047] S2. Decomposing and reconstructing the original vibration signal data using a modal decomposition algorithm to obtain a plurality of reconstructed components.

[0048] S3. Perform data feature hierarchical clustering on the plurality of reconstruction components to obtain a reconstructed aggregate signal.

[0049] This embodiment uses a fully adaptive noise ensemble empirical mode decomposition algorithm (AGG-CEEMDAN, referred to as AGCEEMDAN) connected by an aggregate hierarchical clustering algorithm to reconstruct the time domain data and hierarchical cluster the original vibration signal data of the pole-mounted circuit breaker in different states. The traditional fully adaptive noise ensemble empirical mode decomposition algorithm (CEEMDAN) is difficult to adjust the parameters, requiring empirical adjustment or trial and error, and the calculation speed is slow. It may have modal aliasing problems during operation, and different modes affect each other, resulting in a decrease in the accuracy of the decomposition results. The AGG (Agglomerative Hierarchical Clustering) algorithm is a hierarchical clustering algorithm that constructs a clustering hierarchy in a bottom-up manner. It has diversity in the selection of data cluster spacing and connection criteria, and will produce a hierarchical structure that can effectively aggregate data with the same hierarchical structure. Combining the AGG algorithm with CEEMDAN can just solve the problem of modal aliasing of traditional CEEMDAN. The number of hierarchical clustering is set so that no matter how many reconstructed signals there are after CEEMDAN decomposition, the feature dimensions and data volume after hierarchical clustering are the same, thus avoiding the subsequent large-scale data debugging problems caused by different data dimensions.

[0050] This embodiment completes the above reconstruction and hierarchical clustering process based on the time-domain hierarchical clustering and feature expansion technology of AGCEEMDAN. The steps of the time-domain feature expansion technology based on the AGCEEMDAN algorithm are as follows:

[0051] Step A1: Use CEEMDAN to decompose the original vibration signal of the pole-mounted circuit breaker and reconstruct the one-dimensional signal into multiple reconstructed components IMF m (t).

[0052] Step A2: After decomposition, the IMF m Based on the (t) component, the AGG algorithm module is run, which regards each signal sample as a cluster, uses the Chebyshev distance to calculate the distance between clusters, and uses the maximum distance between clusters (full chain similarity) to measure the reconstructed signal IMF m The similarity between a random data sample in (t) and its adjacent data is that the larger the inter-cluster distance, the smaller the similarity, and vice versa. Then, the hierarchical clustering of each reconstructed signal is performed. The premise of the AGG distance judgment principle is that it takes the IMF m(t) That is, each sample in each reconstructed component after decomposition and reconstruction is regarded as a cluster. Maximum distance between clusters: As an indicator for judging hierarchical clustering between clusters, the larger the distance between clusters, the smaller the similarity, and vice versa. (The similarity calculation method uses the full-chain similarity calculation, that is, the MAX maximum cluster distance), the Chebyshev distance measurement rule is shown in formula (1), and the similarity measurement rule is shown in formula (2):

[0053] d(X,Y)=max|X xy -Y xy | (1);

[0054] d(μ,v)=max(dist(μ[i],v[j])) (2);

[0055] In the formula, x and y are the time and frequency information corresponding to a random signal sample, X and Y are two data samples, and d(X,Y) is the Chebyshev distance between two clusters. (μ,v)∈IMF m (t) are clusters (single samples) at two different positions in each IMF signal, i and j are the relative positions of the two clusters, μ[i] and v[j] represent cluster μ at position i and cluster v at position j, respectively, and d(μ,v0) represents the maximum distance between the two clusters.

[0056] Generally speaking, the data features of the several reconstruction components are hierarchically clustered to obtain a reconstructed aggregate signal, which specifically includes:

[0057] Each of the reconstructed components is regarded as an initial cluster.

[0058] The Chebyshev distance between every two clusters is calculated to obtain the inter-cluster distance.

[0059] The reconstructed components are hierarchically clustered using a complete linkage method, ie, based on the maximum distance between clusters, wherein the maximum distance between clusters is the maximum value of all Chebyshev distances.

[0060] According to the inter-cluster distance and the inter-cluster maximum distance, the clusters with the closest distance are merged until the clustering termination condition is met, and finally a reconstructed aggregate signal is obtained.

[0061] The clustering process is performed when running the AGG module. At each step of hierarchical clustering, the algorithm evaluates the similarity between clusters, uses the Chebyshev distance to calculate the distance between two clusters, and uses the maximum distance as the metric of similarity (as shown in equations (1) and (2)). This process is a dynamic process, and boundary conditions are set for it. When the boundary conditions are met, the algorithm stops. Its main purpose is to allow the features of the signal to be effectively represented and extracted at different levels.

[0062] Step A3: Finally, when the boundary condition M = 3 (the number of hierarchical clusters, i.e., three reconstructed aggregate signals with hierarchical structures are finally output), the algorithm ends, and reconstructed aggregate signals with different hierarchical structure information are generated (co-IMF m＝(1,2,3) (t)), realizes the time domain feature expansion of single-dimensional data, which can be used as the input of subsequent data secondary reconstruction and feature extraction model.

[0063] This signal decomposition algorithm (AGCEEMDAN) does not contain a cyclic module, which is included in CEEMDAN. It only expands the time domain information of the original signal through the inherent empirical decomposition function of the CEEMDAN algorithm. In order to further improve the characteristics of the time domain information (using ZOA-VMD for signal feature extraction), the AGG algorithm is connected to the original signal after CEEMDAN decomposition, and the various reconstructed components after empirical decomposition are aggregated into 3 (M=3) reconstructed aggregate signals CO-IMF with obvious feature differences and expanded time domain information compared to the original signal, so that its hierarchical characteristics are obvious and serve as the input of the next point.

[0064] S4. Use a first variational mode decomposition algorithm to extract features from the reconstructed aggregate signal to obtain a high-feature reconstructed signal; the first variational mode decomposition algorithm is a variational mode decomposition algorithm after parameter optimization using a zebra algorithm.

[0065] This embodiment uses the zebra optimization algorithm (ZOA) to optimize the parameters of the variational mode decomposition algorithm (VMD). The zebra optimization algorithm (ZOA) is used to optimize the structural parameters of the variational mode decomposition algorithm (VMD). The search range and objective function of the zebra population are determined, and the optimal eigenmode number and quadratic penalty factor (k, a) parameters of VMD are determined by continuous iteration to obtain a feature extraction technology based on ZOA-VMD.

[0066] VMD variational mode decomposition is an adaptive, fully recursive mode variation and signal processing method, which performs well in processing data time domain information, but there are still some problems. Including complex manual parameter adjustment process, overfitting problem and high technical calculation requirements. In order to overcome the above problems, this embodiment uses the zebra optimization algorithm ZOA to optimize the VMD algorithm parameters. ZOA uses its unique population defense and attack strategy to automatically optimize the important parameters of VMD locally and globally, which reduces the burden of manual parameter adjustment and further improves the algorithm model operation efficiency. The optimized optimal parameters are more adaptable and reduce the risk of overfitting. In general, the introduction of the ZOA algorithm further enhances the operating efficiency, data processing and feature extraction capabilities of the VMD algorithm, which is more efficient and reliable than the original algorithm.

[0067] The feature extraction process based on ZOA-VMD is as follows:

[0068] Step B1: Establish the VMD algorithm model and initialize the model parameters.

[0069] Step B2: Initialize the zebra population and set parameters such as upper and lower bounds and number of iterations.

[0070] Step B3: The complexity of each reconstructed aggregate signal is measured by sample entropy, and the sample entropy is used as the fitness indicator of the ZOA algorithm. The fitness function is used to evaluate each zebra individual, with the goal of minimizing the mean sample entropy. Each zebra individual represents a possible solution. During the optimization process, the zebra individual will update its position through the strategy to change the corresponding optimization parameter value. Among them, the parameter value refers to the parameters α and K that need to be optimized in VMD.

[0071] The objective function F(t) of the zebra optimization algorithm is shown below, which is used to optimize the combination of evaluation parameters α and K in the VMD algorithm.

[0072]

[0073]

[0074] Among them, F(t) is the fitness function for evaluating the performance of VMD parameters, E i is the sample entropy of each analytical signal, and K is a parameter in the variational mode decomposition (VMD) algorithm, which represents the number of modes of the signal to be decomposed, that is, the number of independent components (or modes) into which the signal is decomposed. i To reconstruct the analytical signal obtained by Hilbert transform of the aggregate signal, H i (k) and H i (j) represents the kth and jth data points of the ith mode, m is the number of consecutive data points considered when calculating the sample entropy, and j is the sample entropy position information in the sequence. For each j position and the range of m points from j+m to j+2m-1 at that position, the average logarithm of the distance between m consecutive points and points from j to j+m-1 is calculated, and finally the sample entropy E of all modal functions is obtained. i The average value is used as the output of the fitness function to optimize the parameter performance in the algorithm.

[0075] Step B4: Set the zebra foraging strategy and the attack and defense strategy in two stages. Calculate and update the optimal position of each zebra through the selection of strategy until the conditions are met and the algorithm terminates. If not, return to step B2 to continue iteration.

[0076] In the zebra optimization algorithm, its foraging strategy is as follows:

[0077]

[0078] Among them, xi is the position information of the zebra individual in the search space, which is a dim-dimensional vector. rand(dim) randomly generates a dim-dimensional random vector in the interval [0,1). Each element of the vector is independent and evenly distributed, which increases the randomness of the algorithm and makes the exploration process more diverse. PZ is the position of the global optimal solution, and I is a random integer between [1,3], which is used to simulate the mutual influence between zebra individuals. Then, the difference between the global optimal solution and the individual position adjusted by the transformation is PZ-I·x i To represent the direction and distance that individual i needs to approach the optimal position. Then add the randomly weighted difference vector to the position x of individual i i Get the updated position Then calculate the new fitness And compare the relationship between the new and old fitness. If This means that the zebra position is optimized and the position is updated to like This means that the original location is better and no update is needed.

[0079] Its offensive and defensive strategies are:

[0080]

[0081] Among them, P s Select parameters for the zebra attack and defense strategies, P s When ≤0.5, an offensive strategy is adopted; P s When R = 0.01, t and T are the number of training iterations and the maximum number of iterations, respectively. AZ is the randomly selected zebra position, simulating a zebra attacked by a predator. After considering the interaction with other zebras and the need to escape from predators, the zebra position information is updated. This stage prevents all zebras from gathering in the same part of the search space, thereby improving the algorithm's global search ability and avoiding falling into local optimal solutions. Its fitness function update mode is the same as the foraging strategy.

[0082] After updating and iteration, the optimal zebra position in the zebra population is:

[0083]

[0084] Step B5: Use the optimal zebra position to optimize the intrinsic mode number and quadratic penalty factor (K, α) parameters of VMD.

[0085] Among them, the process of optimizing the variational mode decomposition algorithm parameters according to the optimal zebra position and the optimal fitness index in this embodiment specifically includes: optimizing the parameters (K, α) of VMD through the optimal zebra position, and the evaluation criterion is the optimal fitness index, that is, the optimal fitness index is used for evaluation, the optimal zebra position is optimized, and the parameters (K, α) of VMD are optimized to appropriate values.

[0086] The optimal position is used for VMD parameter optimization, and the optimal fitness is used for VMD parameter optimization evaluation.

[0087] The optimized source code is: K = int (out_xposbest [1]);

[0088] alpha=int(out_xposbest[0]).

[0089] Among them, out_xposbest is the optimal position after zebra optimization, [0] and [1] represent the first and second parameters after optimization, which are α and K respectively.

[0090] The zebra optimization algorithm returns three parameter variables, namely the optimal position, the optimal objective function, and the fitness. Among them, the optimal position is the best point in the parameter space after zebra optimization, and this position corresponds to the best combination of VMD (k, a). The optimal objective function is the fitness value at the optimal solution, which is used as a criterion to judge whether the VMD parameters are appropriate. Therefore, the zebra optimization algorithm can be directly used to optimize the VMD parameters.

[0091] In general, see Figure 3 The present embodiment provides a method for processing pole-mounted circuit breaker data, which is based on the data feature extraction process of AGCEEMDAN-ZOA-VMD and can be summarized into the following steps:

[0092] Step C1: The collected single-dimensional column-mounted circuit breaker vibration signal is used as the sample input of the model, and the time and frequency domain information after feature extraction is used as the model output.

[0093] Step C2: Use CEEMDAN to reconstruct the time domain information of the input data.

[0094] Step C3: For the reconstructed signal, set the hierarchical parameters and use the AGG algorithm to perform hierarchical clustering of data features.

[0095] Step C4: The reconstructed output signal of AGG-CEEMDAN is used as the input of ZOA-VMD.

[0096] Step C5: Construct the VMD model and initialize the VMD model parameters.

[0097] Step C6: Initialize the zebra population.

[0098] Step C7: Using sample entropy as the fitness index of the ZOA algorithm, determine the objective function of the Banmar optimization algorithm.

[0099] Step C8: Run the zebra optimization algorithm to obtain the optimal zebra position and the optimal fitness index after the foraging stage and the attack and defense stage against predators.

[0100] Step C9: Use the trained zebra optimization algorithm to optimize the (K, α) parameters of VMD.

[0101] Step C10: Run the ZOA-VMD algorithm and then output the reconstructed IMFs signals with high-intensity features.

[0102] Step C11: Take each feature sample processed by the AGCEEMDAN-ZOA-VMD algorithm as input, use Fourier transform (FFT), and then output each spectrum data corresponding to each IMF signal to improve its time and frequency domain characteristics as input for subsequent data classification, prediction and other work.

[0103] This embodiment aims at the problems of difficult testing of pole-mounted circuit breaker data samples, lack of features, small feature quantity, and difficulty in identification caused by small samples. A feature extraction method based on AGCEEMDAN-ZOA-VMD is proposed.

[0104] For the existing small data samples of pole-mounted circuit breakers, if they are directly used as input for data classification and prediction, they may cause inaccurate recognition and poor prediction values ​​due to the small amount of data, thus affecting the reliability of a certain project or equipment. Figure 4 As shown, this embodiment expands the horizontal dimension of the data by extracting deep-level features in the time and frequency domains of the data to ensure the characteristic properties of each state. A single vibration signal input is expanded to three signals with a hierarchical structure through AGCEEMDAN, and the further feature extraction of these three sub-signals through ZOA-VMD makes their features obvious and further expands their time and frequency domain information. The present invention transforms data from a single signal into sub-signals with multiple dimensions and obvious characteristics, so that when performing subsequent tasks such as data recognition, the time and frequency domain features of sub-signals of different dimensions can be identified to make up for the disadvantage of small vertical data volume.

[0105] In order to verify the performance of the algorithm proposed in this embodiment, a vibration sensor is used on the pole-mounted circuit breaker to measure the vibration signals of different states. The data sample entropy is used as the algorithm fitness measurement indicator. The number of iterations is set to 20 and the population size is 20. The measured vibration signals are used as inputs to the AGCEEMDAN-GA-VMD, AGCEEMDAN-PSO-VMD, AGCEEMDAN-WOA-VMD, AGCEEMDAN-GWO-VMD, and AGCEEMDAN-ZOA-VMD models respectively. The algorithm performance is verified by the fitness values ​​after convergence of each algorithm. The results are shown in Table 1.

[0106] Table 1 Comparison of fitness of each algorithm after convergence

[0107]

[0108] It can be seen from Table 1 that compared with the other four algorithm models, the AGCEEMDAN-ZOA-VMD feature extraction technology proposed in this embodiment has the lowest fitness after iterative convergence, which means that compared with other algorithms, this algorithm achieves error minimization and can find the optimal or near-optimal solution, thereby achieving feature enhancement and having better data processing and feature extraction capabilities.

[0109] In another exemplary embodiment of the present application, a method for processing data of a pole-mounted circuit breaker is provided, and the specific steps thereof include:

[0110] 1000 sets of vibration signal data are used as the input of the model. After the data is reconstructed and clustered by CEEMDAN-AGG, it is transformed into co-IMFs (aggregate sub-signals) with obvious hierarchical features to expand the time domain features of the samples and avoid errors in subsequent classification tasks due to insufficient sample features. Co-IMFs are used as the input of the VMD algorithm model, and the ZOA algorithm is used to optimize the (K, α) parameters of VMD based on the entropy of the reconstructed signal samples. The optimal (K, α) is used to deeply extract the potential features of each co-IMFs, followed by Fourier transform to extract the frequency domain features of the data, so as to enhance the time and frequency domain features of the data on the basis of the expanded feature samples. Finally, an output signal with multiple feature dimensions and high time and frequency domain features is obtained.

[0111] Since CEEMDAN is an empirical decomposition algorithm, the CEEMDAN function can be directly called to reconstruct the input data. After reconstruction, the time domain characteristics of the signal are significantly expanded, and the original 1-dimensional acceleration signal is transformed into an 8-dimensional IMF signal. Through the sample entropy distribution in each data, the hierarchical parameter M=3 is set, and the AGG algorithm is used to hierarchically cluster each signal. After clustering each signal, three co-IMFs signals with obvious hierarchical characteristics are generated.

[0112] Initialize VMD parameters, set initial condition parameter init = 1, tolerance parameter tol = 1e-7; set its input to co-IMFs; set its sample entropy calculation function, overlap length m = 2, and then return the mean entropy S of each signal sample; set the ZOA function, set its fitness value to S, specify and initialize the upper and lower bounds, number of iterations, dimension and other information; set the zebra optimization two-stage strategy function, set its return value to the optimal position of a single zebra, the optimal value of the objective function and the transient transformation array of S; set the ZOA algorithm model parameters, as shown in Table 2.

[0113] Table 2 ZOA algorithm model parameters

[0114]

[0115] Call the ZOA function to optimize VMD, and then output the high-feature sub-signals pro-IMFs and fitness values ​​after VMD decomposition. As shown in the table above, with the same parameter settings, compared with the other four algorithm models, the AGCEEMDAN-ZOA-VMD feature extraction technology proposed in this embodiment has the lowest fitness after iterative convergence, which can reflect that the algorithm proposed in this embodiment has better data processing and feature extraction capabilities.

[0116] The CEEMDAN-AGG data processing flow provided in this embodiment uses CEEMDAN to perform empirical mode decomposition of data, and then uses the AGG algorithm to perform hierarchical clustering to expand the time domain characteristics of the data and generate data of the same dimension with hierarchical characteristics for subsequent processing.

[0117] The ZOA algorithm in this embodiment optimizes the parameters and optimization process of VMD, including determining the optimization objective function, and the zebra position optimization VMD strategy, ZOA algorithm model parameters and other settings. The algorithm compares the fitness after optimization. This embodiment proposes a small sample data feature extraction technology based on CEEMDAN-AGG-ZOA-VMD to solve the problem of insufficient data features caused by too small a data volume, including the overall data processing process, input and output conversion between algorithms, and data feature dimension transformation.

[0118] In an exemplary embodiment, a computer device is provided. The computer device may be a server or a terminal. The internal structure diagram thereof may be as follows: Figure 5As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a method for processing data of a pole-mounted circuit breaker is implemented.

[0119] Those skilled in the art will understand that Figure 5 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0120] In an exemplary embodiment, a computer device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above-mentioned method embodiments when executing the computer program.

[0121] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0122] In an exemplary embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0123] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0124] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0125] The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on blockchain, etc., but is not limited thereto. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but is not limited thereto.

[0126] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0127] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A method for processing data of a pole mounted circuit breaker, characterized in that: The pole-mounted circuit breaker data processing method comprises: Obtaining raw vibration signal data of the pole-mounted circuit breaker; Decomposing and reconstructing the original vibration signal data using a modal decomposition algorithm to obtain a plurality of reconstructed components; Performing data feature hierarchical clustering on the plurality of reconstruction components to obtain a reconstructed aggregate signal; A first variational mode decomposition algorithm is used to extract features of the reconstructed aggregate signal to obtain a high-feature reconstructed signal; the first variational mode decomposition algorithm is a variational mode decomposition algorithm after parameter optimization using a zebra algorithm.

2. The method for processing data of a pole mounted circuit breaker according to claim 1, characterized in that: The process of optimizing the parameters of the variational mode decomposition algorithm using the zebra algorithm includes: With the goal of minimizing the mean value of sample entropy, the fitness function is used to evaluate each zebra individual in the zebra algorithm to obtain the optimal zebra position and the optimal fitness index; The variational mode decomposition algorithm parameters are optimized according to the optimal zebra position and the optimal fitness index.

3. The method for processing data of a pole mounted circuit breaker according to claim 1, characterized in that: Performing data feature hierarchical clustering on the plurality of reconstruction components to obtain a reconstructed aggregate signal specifically includes: Consider each of the reconstructed components as an initial cluster; Calculate the inter-cluster distance between every two initial clusters; hierarchical clustering is performed on the reconstructed components based on the maximum inter-cluster distance; the maximum inter-cluster distance is the maximum value of all the inter-cluster distances; According to the inter-cluster distance and the inter-cluster maximum distance, the clusters with the closest distance are merged to obtain a reconstructed aggregate signal.

4. The method for processing data of a pole mounted circuit breaker according to claim 3, characterized in that: The calculating of the inter-cluster distance between every two initial clusters specifically includes: The Chebyshev distance between each two initial clusters is calculated to obtain the inter-cluster distance.

5. The method for processing data of a pole mounted circuit breaker according to claim 1, characterized in that: After extracting features from the reconstructed aggregate signal using the first variational mode decomposition algorithm to obtain a high-feature reconstructed signal, the method further includes: Perform Fourier transform on the high-feature reconstructed signal to obtain spectrum data.

6. The method for processing data of a pole mounted circuit breaker according to claim 1, characterized in that: Performing data feature hierarchical clustering on the plurality of reconstruction components to obtain a reconstructed aggregate signal specifically includes: The data features of the plurality of reconstruction components are hierarchically clustered using an aggregation hierarchical clustering algorithm to obtain a reconstructed aggregation signal.

7. The method for processing data of a pole mounted circuit breaker according to claim 1, characterized in that: The objective function of the zebra algorithm is: Among them, E i For each analytic signal, the sample entropy, H i To reconstruct the analytical signal obtained by Hilbert transform of the aggregate signal, m is the number of continuous data points considered when calculating the sample entropy, and j is the sample entropy position information in the sequence.

8. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the pole-mounted circuit breaker data processing method according to any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the pole-mounted circuit breaker data processing method according to any one of claims 1 to 7 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the pole-mounted circuit breaker data processing method according to any one of claims 1 to 7 is implemented.

Images