Insulator classification method, device, equipment and storage medium

Abstract

The present invention discloses an insulator classification method, device, equipment and storage medium, the method comprising: obtaining target sound wave frequency data and initial parameters of an echo state network; the initial parameters include initial leakage parameters, initial input scaling parameters, initial input weight matrix, initial circulation weight matrix and initial output weight matrix; adjusting the initial leakage parameters and initial input scaling parameters to obtain target leakage parameters and target input scaling parameters; adjusting the initial input weight matrix according to the target input scaling parameters to obtain the target input weight matrix: based on the echo state network, according to the target sound wave frequency data, target leakage parameters, target input weight matrix, initial circulation weight matrix and initial output weight matrix, determining the classification category of the insulator. The embodiments of the present invention improve the detection accuracy of the insulator and reduce the detection cost of the insulator.

Description

Technical Field

[0001] The present invention relates to the technical field of insulator detection, and in particular to an insulator classification method, device, equipment and storage medium. Background Art

[0002] Insulators are indispensable insulating components in power transmission lines. Their operating conditions affect the safety of power grid operation and the reliability of power supply. Timely detection and replacement of faulty insulators (i.e. perforated insulators) in the power grid can improve the safety of power grid operation and the reliability of power supply.

[0003] Traditional insulator detection technologies, such as those based on support vector machines and those based on multi-layer perceptron networks, have low detection accuracy. In order to improve the detection accuracy of insulators, it is necessary to spend a lot of money to hire professional maintenance personnel to detect insulators one by one, which is costly. Summary of the invention

[0004] The present invention provides an insulator classification method, device, equipment and storage medium to improve the detection accuracy of insulators and reduce the detection cost of insulators.

[0005] According to one aspect of the present invention, there is provided an insulator classification method, comprising:

[0006] Obtaining target sound wave frequency data and initial parameters of the echo state network; the initial parameters include an initial leakage parameter, an initial input scaling parameter, an initial input weight matrix, an initial circulation weight matrix, and an initial output weight matrix;

[0007] Adjusting an initial leakage parameter and an initial input scaling parameter to obtain a target leakage parameter and a target input scaling parameter;

[0008] According to the target input scaling parameter, adjust the initial input weight matrix to obtain the target input weight matrix:

[0009] Based on the echo state network, the classification category of the insulator is determined according to the target acoustic wave frequency data, the target leakage parameter, the target input weight matrix, the initial circulation weight matrix and the initial output weight matrix.

[0010] According to another aspect of the present invention, there is provided an insulator classification device, comprising:

[0011] A data acquisition module is used to acquire target sound wave frequency data and initial parameters of the echo state network; the initial parameters include initial leakage parameters, initial input scaling parameters, initial input weight matrix, initial cycle weight matrix and initial output weight matrix;

[0012] A parameter determination module, used for adjusting an initial leakage parameter and an initial input scaling parameter to obtain a target leakage parameter and a target input scaling parameter;

[0013] The matrix determination module is used to adjust the initial input weight matrix according to the target input scaling parameters to obtain the target input weight matrix:

[0014] The category determination module is used to determine the classification category of the insulator based on the echo state network according to the target sound wave frequency data, the target leakage parameter, the target input weight matrix, the initial cycle weight matrix and the initial output weight matrix.

[0015] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0016] at least one processor; and

[0017] a memory communicatively connected to at least one processor; wherein,

[0018] The memory stores a computer program that can be executed by at least one processor. The computer program is executed by the at least one processor so that the at least one processor can execute the insulator classification method according to any embodiment of the present invention.

[0019] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the insulator classification method according to any embodiment of the present invention when executed.

[0020] The technical solution of the embodiment of the present invention is to obtain target sound wave frequency data and initial parameters of the echo state network; the initial parameters include initial leakage parameters, initial input scaling parameters, initial input weight matrix, initial circulation weight matrix and initial output weight matrix; adjust the initial leakage parameters and initial input scaling parameters to obtain target leakage parameters and target input scaling parameters; adjust the initial input weight matrix according to the target input scaling parameters to obtain the target input weight matrix: based on the echo state network, determine the classification category of the insulator according to the target sound wave frequency data, target leakage parameters, target input weight matrix, initial circulation weight matrix and initial output weight matrix. The above technical solution obtains an echo state network that is in the optimal state after training by adjusting the initial leakage parameters, initial input scaling parameters and initial input weight matrix in the echo state network; the target acoustic wave frequency data is classified using the above echo state network to obtain more accurate insulator classification categories, which is convenient for more accurate classification of insulators detected in the power grid, and is convenient for accurate identification of faulty insulators (i.e., perforated insulators) in the power grid, thereby improving the detection accuracy of insulators, reducing the use of human resources, and reducing the cost of hiring additional professional maintenance personnel to perform one-by-one inspections of insulators in the power grid, thereby reducing the detection cost of insulators.

[0021] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0023] Figure 1 is a flow chart of an insulator classification method provided according to Embodiment 1 of the present invention;

[0024] Figure 2 is a flow chart of an insulator classification method provided according to the second embodiment of the present invention;

[0025] Figure 3 is a structural schematic diagram of an insulator classification device provided according to Embodiment 3 of the present invention;

[0026] Figure 4 It is a schematic diagram of the structure of an electronic device for implementing the insulator classification method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0027] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0028] It should be noted that the terms "target" and "initial" and the like in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0029] In addition, it should be noted that the collection, storage, use, processing, transmission, provision and disclosure of target sound wave frequency data, etc. involved in the technical solution of the present invention are in compliance with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0030] Embodiment 1

[0031] Figure 1 This is a flow chart of an insulator classification method provided in the first embodiment of the present invention. This embodiment is applicable to the case of repairing faulty insulators in a power grid. The method can be executed by an insulator classification device, which can be implemented in the form of hardware and / or software and can be configured in an electronic device, which can be a mobile phone, a laptop computer, a desktop computer, etc. Figure 1 As shown, the method includes:

[0032] S101, obtaining target sound wave frequency data and initial parameters of the echo state network; the initial parameters include initial leakage parameters, initial input scaling parameters, initial input weight matrix, initial circulation weight matrix and initial output weight matrix.

[0033] Among them, the target acoustic wave frequency data may refer to the acoustic wave frequency data obtained by capturing the acoustic wave frequency signal of the insulator to be detected. The insulator to be detected may refer to the insulator that needs to be detected in the power grid. The initial parameters may refer to the parameters in the echo state network before the echo state network is trained. The initial parameters can be pre-set according to actual needs or randomly generated. The initial parameters of the echo state network include an initial leakage parameter, an initial input scaling parameter, an initial input weight matrix, an initial loop weight matrix, and an initial output weight matrix. Among them, the initial leakage parameter (denoted as α 1 ) takes values ​​between [0,1], that is, α 1 ∈[0,1], the initial input weight matrix (denoted as W i 1 n ), the initial cycle weight matrix (denoted as W) and the initial output weight matrix (denoted as W out ) are all real numbers, that is, W i 1 n ∈R,W∈R,W out ∈R.

[0034] Specifically, an ultrasonic detector is used to capture the acoustic frequency signal of the insulator in the power grid to obtain the target acoustic frequency data; and the initial parameters of the echo state network are pre-set according to actual needs, or the initial parameters of the echo state network are randomly generated. The initial parameters of the echo state network include initial leakage parameters, initial input scaling parameters, initial input weight matrix, initial circulation weight matrix and initial output weight matrix.

[0035] Exemplarily, target sound wave frequency data is obtained from a pre-configured database; and initial parameters of the echo state network are pre-set according to actual needs.

[0036] Exemplarily, the detection environment of insulators in the power grid is simulated in the laboratory. When the phase-to-ground voltage in the laboratory is 7.95 kilovolts (i.e., 7.95kV) and the phase-to-ground voltage of the distribution system in the laboratory is 13.8kV, an ultrasonic detector with an ultrasonic frequency of 500 kilohertz (i.e., 500kHz) is used in a pre-configured acrylic cavity at a distance of 0.4 meters from the insulator to capture the acoustic frequency signal of the insulator within 50 seconds to obtain the target acoustic frequency data; and the initial parameters of the echo state network are randomly generated. It should be noted that the target acoustic frequency data includes the acoustic frequency signals of at least 50 insulators to ensure that the acoustic frequency signals of insulators above 10kHz can be captured in each ultrasonic cycle. Among them, one ultrasonic cycle can be determined by the following formula:

[0037]

[0038] Among them, f is the ultrasonic frequency, and T is the ultrasonic period. It should be noted that the phase-to-ground voltage in the laboratory is 7.95 kilovolts (i.e., 7.95 kV), and the phase-to-ground voltage of the power distribution system in the laboratory is 13.8 kV. This is to ensure that the voltage is consistent with the voltage in the insulators exposed in the power grid to simulate the voltage of the insulators in the power grid.

[0039] S102: Adjust the initial leakage parameter and the initial input scaling parameter to obtain a target leakage parameter and a target input scaling parameter.

[0040] The target leakage parameter may refer to a leakage parameter obtained after adjusting the initial leakage parameter. The target input scaling parameter may refer to an input scaling parameter obtained after adjusting the initial input scaling parameter.

[0041] Specifically, the initial leakage parameter and the initial input scaling parameter in the echo state network are input into a preset mesh adaptive direct search (MADS) algorithm, and the target leakage parameter and the target input scaling parameter are obtained after adjustment by the algorithm. The preset MADS algorithm may be one of a quadratic programming mesh adaptive direct search (QPMADS) algorithm, a lower triangular mesh adaptive direct search (LTMADS) algorithm, an orthogonal mesh adaptive direct search (OrthoMADS) algorithm, and an orthogonal triangular decomposition mesh adaptive direct search (QRMADS) algorithm. The MADS algorithm is a pattern search algorithm that calculates a series of points near an optimal point, and is used to search for an optimal point in a variable space.

[0042] Exemplarily, the initial leakage parameter and the initial input scaling parameter in the echo state network are input into the QPMADS algorithm, and after adjustment by the algorithm, the target leakage parameter and the target input scaling parameter are obtained.

[0043] Exemplarily, the initial leakage parameter and the initial input scaling parameter in the echo state network are input into the OrthoMADS algorithm, and after adjustment by the algorithm, the target leakage parameter and the target input scaling parameter are obtained.

[0044] S103 . Adjust the initial input weight matrix according to the target input scaling parameter to obtain the target input weight matrix.

[0045] The target input weight matrix may refer to an input weight matrix obtained after adjusting the initial input weight matrix.

[0046] Specifically, the ratio of the initial input weight matrix to the spectral radius of the echo state network reserve layer is calculated, and the product of the ratio and the target input scaling parameter is used as the target input weight matrix. Accordingly, the target input weight matrix can be determined by the following formula:

[0047]

[0048] in, is the initial input weight matrix, λ max is the spectral radius of the echo state network reserve layer (i.e., the maximum value of each initial input weight matrix in the echo state network reserve layer), γ is the target input scaling parameter, Enter the weight matrix for the target, R is a real number.

[0049] S104. Based on the echo state network, the classification category of the insulator is determined according to the target sound wave frequency data, the target leakage parameter, the target input weight matrix, the initial circulation weight matrix and the initial output weight matrix.

[0050] Among them, the classification categories of insulators may include pollution-free and non-perforated insulators, pollution-free and perforated insulators, pollution-free and non-perforated insulators, pollution-free and perforated insulators, pollution-free and perforated insulators, non-perforated insulators with impurities (salt addition), and porous wet insulators. It should be noted that the classification categories of insulators can be pre-set in the echo state network, or obtained by pre-training the echo state network, without specific restrictions. For example, 200 classification categories of insulators are pre-set in the echo state network, and the echo state network compares the input target sound wave frequency data with the classification categories of these 200 insulators one by one, and then classifies the sound wave frequency signals in the target sound wave frequency data into the classification categories of these 200 insulators. In the case where there are 200 classification categories of insulators, the classification categories of insulators obtained by the echo state network are the most accurate.

[0051] Specifically, based on the echo state network, according to the target acoustic wave frequency data, the target leakage parameter, the target input weight matrix, the initial cycle weight matrix and the initial output weight matrix, the insulator classification and recognition model is determined; the target acoustic wave frequency data is classified using the insulator classification and recognition model to obtain the classification category of the insulator. Among them, the insulator classification and recognition model is used to classify the insulators. It can be understood that the classification category of the insulators obtained based on the echo state network not only takes into account the two characteristics of the insulators in the power grid (i.e., insulators without perforations and impurities (salt addition) and porous wet insulators), but also can distinguish between contaminated insulators and perforated insulators in the power grid, so as to facilitate the timely detection and replacement of perforated insulators in the power grid, so as to improve the safety of the power grid operation and the reliability of the power supply of the power grid.

[0052] Optionally, according to the target sound wave frequency data, the target leakage parameter, the target input weight matrix and the initial cycle weight matrix, the neuron activation vector in the echo state network reserve layer is determined; according to the target sound wave frequency data, the neuron activation vector and the initial output weight matrix, the insulator classification and recognition model is determined; the target sound wave frequency data is classified using the insulator classification and recognition model to obtain the classification category of the insulator.

[0053] It should be noted that the echo state network includes an input layer, a reserve layer and an output layer. The input layer is used to input target sound wave frequency data, the reserve layer is used to generate a complex dynamic space that changes continuously with the target sound wave frequency data input by the input layer, and the output layer is used to output the insulator classification and recognition model.

[0054] Specifically, the target sound wave frequency data is recorded as u(n), u(n)∈R, R is a real number, and n is a positive integer; the target leakage parameter is recorded as α 2 , α 2 ∈[0,1]; the target input weight matrix is ​​recorded as The initial cycle weight matrix is ​​recorded as W, W∈R; the neuron activation vector in the echo state network reserve layer is recorded as x(n), x(n)∈R. Correspondingly, the neuron activation vector in the echo state network reserve layer (i.e., x(n)) can be determined by the following formula:

[0055]

[0056] Where n represents the nth sample sound wave frequency data in the target sound wave frequency data, is the update of x(n), and tanh() is the activation function of the echo state network reserve layer. Then, according to the target sound wave frequency data u(n), the neuron activation vector x(n) in the echo state network reserve layer, and the initial output weight matrix, the insulator classification and recognition model is determined by the following formula:

[0057] y(n)=W out [1;u(n);x(n)];

[0058] Among them, y(n) is the insulator classification and recognition model, y(n)∈R, R is a real number, W out is the initial output weight matrix, W out ∈R. Then, the insulator classification and recognition model is used to classify the target acoustic wave frequency data to obtain the classification category of the insulator, and the insulators in the power grid are divided into categories such as pollution-free and non-perforated insulators, pollution-free and perforated insulators, pollution-free and non-perforated insulators, pollution-free and perforated insulators, pollution-free and perforated insulators, non-perforated insulators with impurities (salt added) and porous wet insulators, so as to detect the perforated insulators in the power grid according to the classification category of the insulators, accurately identify the perforated insulators in the power grid, and then facilitate professional maintenance personnel to replace all perforated insulators to ensure the normal operation of the insulators in the power grid, improve the safety of the power grid operation and the reliability of the power supply of the power grid. The above technical scheme provides a specific method for determining the classification category of insulators based on the echo state network, so that the determination of the classification category of insulators has a certain regularity.

[0059] The technical solution of the embodiment of the present invention is to obtain target sound wave frequency data and initial parameters of the echo state network; the initial parameters include initial leakage parameters, initial input scaling parameters, initial input weight matrix, initial circulation weight matrix and initial output weight matrix; adjust the initial leakage parameters and initial input scaling parameters to obtain target leakage parameters and target input scaling parameters; adjust the initial input weight matrix according to the target input scaling parameters to obtain the target input weight matrix: based on the echo state network, determine the classification category of the insulator according to the target sound wave frequency data, target leakage parameters, target input weight matrix, initial circulation weight matrix and initial output weight matrix. The above technical solution obtains an echo state network that is in the optimal state after training by adjusting the initial leakage parameters, initial input scaling parameters and initial input weight matrix in the echo state network; the target acoustic wave frequency data is classified using the above echo state network to obtain more accurate insulator classification categories, which is convenient for more accurate classification of insulators detected in the power grid, and is convenient for accurate identification of faulty insulators (i.e., perforated insulators) in the power grid, thereby improving the detection accuracy of insulators, reducing the use of human resources, and reducing the cost of hiring additional professional maintenance personnel to perform one-by-one inspections of insulators in the power grid, thereby reducing the detection cost of insulators.

[0060] Embodiment 2

[0061] Figure 2This is a flow chart of an insulator classification method provided in Example 2 of the present invention. Based on the above example, this example further optimizes "adjusting the initial leakage parameter and the initial input scaling parameter to obtain the target leakage parameter and the target input scaling parameter" and provides an optional implementation scheme. It should be noted that for the parts not described in detail in the example of the present invention, reference can be made to the relevant descriptions of other examples. Figure 2 As shown, the method includes:

[0062] S201, obtaining target sound wave frequency data and initial parameters of the echo state network; the initial parameters include initial leakage parameters, initial input scaling parameters, initial input weight matrix, initial circulation weight matrix and initial output weight matrix.

[0063] S202: Determine a search times threshold according to a preset search direction of the grid adaptive direct search algorithm.

[0064] Among them, the grid adaptive direct search algorithm can be one of the QPMADS algorithm, the LTMADS algorithm, the OrthoMADS algorithm and the QRMADS algorithm. The search direction depends on the grid adaptive direct search algorithm. For example, a relatively dense direction in the variable space of the grid adaptive direct search algorithm is used as a preset search direction. The search number threshold may refer to the maximum number of searches in the grid adaptive direct search algorithm. It should be noted that the grid adaptive direct search algorithm only searches one preset search direction at a time.

[0065] Specifically, the total number of search directions preset by the grid adaptive direct search algorithm is counted, and the total number is used as the search number threshold of the grid adaptive direct search algorithm.

[0066] Exemplarily, if the grid adaptive direct search algorithm is the OrthoMADS algorithm, the total number of search directions preset by the OrthoMADS algorithm is counted (denoted as M), and the search number threshold of the OrthoMADS algorithm is M.

[0067] S203 , using a grid adaptive direct search algorithm to iteratively adjust the initial leakage parameter and the initial input scaling parameter to obtain a target leakage parameter and a target input scaling parameter.

[0068] It should be noted that the grid adaptive direct search algorithm determines the optimal leakage parameter and input scaling parameter in the next search process through the following formula:

[0069]

[0070] Among them, M k is the set of grid points, is the grid size parameter, x is the starting point of the search, Vk is a real number, that is, V k ∈R, k is the kth grid, D z is the search direction, z is the zth search, and z is a positive integer, that is, z∈N. It should be noted that when the grid adaptive direct search algorithm is the OrthoMADS algorithm, D z For n -I n ],I n is the identity matrix of dimension n.

[0071] Optionally, the initial leakage parameters and initial input scaling parameters of the echo state network are input into the grid adaptive direct search algorithm to obtain the leakage parameters and input scaling parameters after the first adjustment; the above leakage parameters and input scaling parameters are used as inputs of the next grid adaptive direct search algorithm; when the number of searches is greater than a search number threshold, the leakage parameters and input scaling parameters corresponding to the search number are used as target leakage parameters and target input scaling parameters.

[0072] Optionally, a grid adaptive direct search algorithm is used to iteratively adjust the initial leakage parameter and the initial input scaling parameter to obtain the target leakage parameter and the target input scaling parameter, which can be: when the current search number is less than or equal to the search number threshold, the current leakage parameter and the current input scaling parameter corresponding to the current search number are input into the echo state network to determine the current classification accuracy corresponding to the current search number; when the current classification accuracy is greater than or equal to the preset target classification accuracy, the current classification accuracy is used as the new target classification accuracy; when the current search number is greater than the search number threshold, the leakage parameter and the input scaling parameter corresponding to the new target classification accuracy are used as the target leakage parameter and the target input scaling parameter.

[0073] The preset target classification accuracy can be randomly set in advance, or can be preset according to the classification accuracy corresponding to the first search of the grid adaptive direct search algorithm. Exemplarily, if the classification accuracy corresponding to the first search of the grid adaptive direct search algorithm is Acc, the preset target classification accuracy can be Acc.

[0074] Specifically, the search number threshold is recorded as M, the current search number is recorded as z, z = 1, 2, ..., M, and the preset target classification accuracy is recorded as Acc 0 Based on the grid adaptive direct search algorithm (such as OrthoMADS algorithm), when z≤M, the current leakage parameter and the current input scaling parameter corresponding to the current search number are input into the echo state network to obtain the current classification accuracy corresponding to the current search number (denoted as Acc 1 ); if Acc 1 ≥Acc 0, then the current classification accuracy Acc 1 As the new target classification accuracy, otherwise, the new target classification accuracy is still Acc 0 ; In the case of z>M, the new target classification accuracy Acc 1 The corresponding leakage parameters and input scaling parameters are used as the target leakage parameters and target input scaling parameters.

[0075] It can be understood that the grid adaptive direct search algorithm is used to iteratively adjust the initial leakage parameters and initial input scaling parameters of the echo state network, and the classification accuracy corresponding to each search of the grid adaptive direct search algorithm is calculated through the echo state network, and the maximum value of the above classification accuracies is used as the target classification accuracy; then the leakage parameters and input scaling parameters corresponding to the target classification accuracy are determined as the target leakage parameters and target input scaling parameters, thereby realizing the adjustment of the initial leakage parameters and initial input scaling parameters of the echo state network, adjusting the initial leakage parameters and initial input scaling parameters of the echo state network to the optimal state, and then obtaining the echo state network in the optimal state after training, so that a more accurate insulator classification category can be obtained based on the echo state network in the future.

[0076] Optionally, when the current search is the first search, a grid adaptive direct search algorithm is used to adjust the initial leakage parameters and the initial input scaling parameters to obtain the leakage parameters and the input scaling parameters after the first processing; the leakage parameters and the input scaling parameters after the first processing are input into the echo state network to obtain the classification accuracy corresponding to the first search, and the classification accuracy corresponding to the first search is used as the preset target classification accuracy.

[0077] Exemplarily, if the current search number is z (z = 1, 2, ..., M), M is the search number threshold M; the OrthoMADS algorithm is used to adjust the leakage parameter and input scaling parameter of the echo state network; in the case of z = 1, the initial leakage parameter and initial input scaling parameter of the echo state network are input into the OrthoMADS algorithm to obtain the leakage parameter and input scaling parameter after the first processing; the leakage parameter and input scaling parameter after the first processing are input into the echo state network to obtain the classification accuracy corresponding to the first search, and the classification accuracy is used as the preset target classification accuracy. A method for determining the preset target classification accuracy is provided, which makes the determination of the preset target classification accuracy more scientific and reasonable.

[0078] S204: Adjust the initial input weight matrix according to the target input scaling parameter to obtain the target input weight matrix.

[0079] S205. Based on the echo state network, determine the classification category of the insulator according to the target sound wave frequency data, the target leakage parameter, the target input weight matrix, the initial circulation weight matrix and the initial output weight matrix.

[0080] The technical solution of the embodiment of the present invention determines the search times threshold according to the search direction preset by the grid adaptive direct search algorithm; the grid adaptive direct search algorithm is used to iteratively adjust the initial leakage parameter and the initial input scaling parameter to obtain the target leakage parameter and the target input scaling parameter. A specific method for adjusting the initial leakage parameter and the initial input scaling parameter in the echo state network is provided, and the grid adaptive direct search algorithm is used to adjust the leakage parameter and the input scaling parameter in the echo state network to the optimal state, thereby obtaining the target leakage parameter and the target input scaling parameter, so as to facilitate the subsequent adjustment of the input weight matrix in the echo state network to the optimal state according to the target leakage parameter and the target input scaling parameter, thereby improving the accuracy of the subsequent insulator classification category obtained based on the echo state network.

[0081] Embodiment 3

[0082] Figure 3 This is a schematic diagram of the structure of an insulator classification device provided in the third embodiment of the present invention. This embodiment is applicable to the case of repairing faulty insulators in a power grid. The device can be implemented in the form of hardware and / or software and can be configured in an electronic device, which can be a mobile phone, a laptop computer, a desktop computer, etc. Figure 3 As shown, the device comprises:

[0083] The data acquisition module 301 is used to acquire the target sound wave frequency data and the initial parameters of the echo state network; the initial parameters include an initial leakage parameter, an initial input scaling parameter, an initial input weight matrix, an initial circulation weight matrix and an initial output weight matrix;

[0084] A parameter determination module 302, used to adjust an initial leakage parameter and an initial input scaling parameter to obtain a target leakage parameter and a target input scaling parameter;

[0085] A matrix determination module 303 is used to adjust the initial input weight matrix according to the target input scaling parameter to obtain the target input weight matrix;

[0086] The category determination module 304 is used to determine the classification category of the insulator based on the echo state network according to the target sound wave frequency data, the target leakage parameter, the target input weight matrix, the initial circulation weight matrix and the initial output weight matrix.

[0087] The technical solution of the embodiment of the present invention obtains target sound wave frequency data and initial parameters of the echo state network through a data acquisition module; the initial parameters include initial leakage parameters, initial input scaling parameters, initial input weight matrix, initial circulation weight matrix and initial output weight matrix; the target leakage parameters and target input scaling parameters are determined through a parameter determination module; the target input weight matrix is ​​determined through a matrix determination module; and the classification category of the insulator is determined through a category determination module based on the echo state network according to the target sound wave frequency data, target leakage parameters, target input weight matrix, initial circulation weight matrix and initial output weight matrix. The above technical solution obtains an echo state network that is in the optimal state after training by adjusting the initial leakage parameters, initial input scaling parameters and initial input weight matrix in the echo state network; the target acoustic wave frequency data is classified using the above echo state network to obtain more accurate insulator classification categories, which is convenient for more accurate classification of insulators detected in the power grid, and is convenient for accurate identification of faulty insulators (i.e., perforated insulators) in the power grid, thereby improving the detection accuracy of insulators, reducing the use of human resources, and reducing the cost of hiring additional professional maintenance personnel to perform one-by-one inspections of insulators in the power grid, thereby reducing the detection cost of insulators.

[0088] Optionally, the category determination module 304 includes:

[0089] An activation vector determination unit, used to determine a neuron activation vector in the echo state network reserve layer according to target sound wave frequency data, target leakage parameters, target input weight matrix, and initial cycle weight matrix;

[0090] The recognition model determination unit is used to determine the insulator classification recognition model according to the target sound wave frequency data, the neuron activation vector and the initial output weight matrix;

[0091] The classification category determination unit is used to classify the target sound wave frequency data using the insulator classification recognition model to obtain the classification category of the insulator.

[0092] Optionally, the parameter determination module 302 includes:

[0093] A number threshold determination unit, used to determine a search number threshold according to a search direction preset by a grid adaptive direct search algorithm;

[0094] The parameter determination unit is used to iteratively adjust the initial leakage parameter and the initial input scaling parameter by using a grid adaptive direct search algorithm to obtain a target leakage parameter and a target input scaling parameter.

[0095] Optionally, the parameter determination unit is specifically used to:

[0096] When the current search times is less than or equal to the search times threshold, the current leakage parameter and the current input scaling parameter corresponding to the current search times are input into the echo state network to determine the current classification accuracy corresponding to the current search times; when the current classification accuracy is greater than or equal to the preset target classification accuracy, the current classification accuracy is used as the new target classification accuracy; when the current search times is greater than the search times threshold, the leakage parameter and the input scaling parameter corresponding to the new target classification accuracy are used as the target leakage parameter and the target input scaling parameter.

[0097] Optionally, the device further comprises:

[0098] A leakage parameter and input scaling parameter determination module, used to adjust the initial leakage parameter and the initial input scaling parameter by using a grid adaptive direct search algorithm when the current search is the first search, so as to obtain the leakage parameter and the input scaling parameter after the first processing;

[0099] The target classification accuracy determination module is used to input the leakage parameter and input scaling parameter after the first processing into the echo state network to obtain the classification accuracy corresponding to the first search, and use the classification accuracy corresponding to the first search as the preset target classification accuracy.

[0100] The insulator classification device provided in the embodiment of the present invention can execute the insulator classification method provided in any embodiment of the present invention, and has corresponding functional modules and beneficial effects for executing each insulator classification method.

[0101] Embodiment 4

[0102] Figure 4 A schematic diagram of the structure of an electronic device 10 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0103] like Figure 4As shown, the electronic device 10 includes at least one processor 11, and a memory connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 to the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, ROM 12 and RAM 13 are connected to each other through a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0104] A number of components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0105] The processor 11 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as the insulator classification method.

[0106] In some embodiments, the insulator classification method may be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the insulator classification method described above may be performed. Alternatively, in other embodiments, the processor 11 may be configured to perform the insulator classification method in any other appropriate manner (e.g., by means of firmware).

[0107] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0108] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0109] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0110] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0111] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0112] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0113] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0114] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for classifying insulators, characterized in that: The method comprises: Acquiring target sound wave frequency data and initial parameters of the echo state network; the initial parameters include initial leakage parameters, initial input scaling parameters, initial input weight matrix, initial circulation weight matrix and initial output weight matrix; Adjusting the initial leakage parameter and the initial input scaling parameter to obtain a target leakage parameter and a target input scaling parameter; According to the target input scaling parameter, adjusting the initial input weight matrix to obtain a target input weight matrix; Based on the echo state network, the classification category of the insulator is determined according to the target sound wave frequency data, the target leakage parameter, the target input weight matrix, the initial circulation weight matrix and the initial output weight matrix.

2. The method according to claim 1, characterized in that The echo state network-based method determines the classification category of the insulator according to the target sound wave frequency data, the target leakage parameter, the target input weight matrix, the initial cycle weight matrix, and the initial output weight matrix, including: Determining a neuron activation vector in an echo state network reserve layer according to the target sound wave frequency data, the target leakage parameter, the target input weight matrix, and the initial cycle weight matrix; Determining an insulator classification and recognition model according to the target sound wave frequency data, the neuron activation vector, and the initial output weight matrix; The insulator classification and recognition model is used to classify the target sound wave frequency data to obtain the classification category of the insulator.

3. The method according to claim 1, characterized in that The adjusting the initial leakage parameter and the initial input scaling parameter to obtain a target leakage parameter and a target input scaling parameter includes: Determine the search number threshold according to the search direction preset by the grid adaptive direct search algorithm; The grid adaptive direct search algorithm is used to iteratively adjust the initial leakage parameter and the initial input scaling parameter to obtain a target leakage parameter and a target input scaling parameter.

4. The method according to claim 3, characterized in that The adopting the grid adaptive direct search algorithm to iteratively adjust the initial leakage parameter and the initial input scaling parameter to obtain the target leakage parameter and the target input scaling parameter includes: When the current search number is less than or equal to the search number threshold, inputting the current leakage parameter and the current input scaling parameter corresponding to the current search number into the echo state network to determine the current classification accuracy corresponding to the current search number; When the current classification accuracy is greater than or equal to a preset target classification accuracy, the current classification accuracy is used as a new target classification accuracy; When the current search times is greater than the search times threshold, the leakage parameter and input scaling parameter corresponding to the new target classification accuracy are used as the target leakage parameter and the target input scaling parameter.

5. The method according to claim 4, characterized in that The method further comprises: In the case where the current search is the first search, the initial leakage parameter and the initial input scaling parameter are adjusted using the grid adaptive direct search algorithm to obtain the leakage parameter and the input scaling parameter after the first processing; The leakage parameter and the input scaling parameter after the first processing are input into the echo state network to obtain the classification accuracy corresponding to the first search, and the classification accuracy corresponding to the first search is used as the preset target classification accuracy.

6. An insulator classification device, characterized in that: include: A data acquisition module, used to obtain target sound wave frequency data and initial parameters of the echo state network; The initial parameters include an initial leakage parameter, an initial input scaling parameter, an initial input weight matrix, an initial cycle weight matrix and an initial output weight matrix; a parameter determination module, used for adjusting the initial leakage parameter and the initial input scaling parameter to obtain a target leakage parameter and a target input scaling parameter; A matrix determination module, used for adjusting the initial input weight matrix according to the target input scaling parameter to obtain a target input weight matrix; The category determination module is used to determine the classification category of the insulator based on the echo state network according to the target sound wave frequency data, the target leakage parameter, the target input weight matrix, the initial cycle weight matrix and the initial output weight matrix.

7. The device according to claim 6, characterized in that The category determination module comprises: an activation vector determination unit, configured to determine a neuron activation vector in an echo state network reserve layer according to the target sound wave frequency data, the target leakage parameter, the target input weight matrix, and the initial cycle weight matrix; A recognition model determination unit, used to determine an insulator classification recognition model according to the target sound wave frequency data, the neuron activation vector and the initial output weight matrix; The classification category determination unit is used to classify the target sound wave frequency data using the insulator classification recognition model to obtain the classification category of the insulator.

8. The device according to claim 6, characterized in that The parameter determination module comprises: A number threshold determination unit, used to determine a search number threshold according to a search direction preset by a grid adaptive direct search algorithm; The parameter determination unit is used to iteratively adjust the initial leakage parameter and the initial input scaling parameter by adopting the grid adaptive direct search algorithm to obtain a target leakage parameter and a target input scaling parameter.

9. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can perform the insulator classification method according to any one of claims 1 to 5.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the insulator classification method according to any one of claims 1 to 5 when executed.

Images