Insulator defect detection method

Abstract

The invention discloses an insulator defect detection method. The method comprises the following steps: acquiring first pre-detection data and second pre-detection data of an insulator; wherein the first pre-detection data and the second pre-detection data are pre-detection data corresponding to different types of pre-detection methods; encoding the first pre-detection data and the second pre-detection data to respectively form a first feature vector and a second feature vector; performing vector coupling on the first feature vector and the second feature vector to form a total vector; and determining the defect state of the insulator according to the total vector. According to the technical scheme, two detection methods are combined, the insulator detection accuracy is improved, the fault risk can be found in time, corresponding maintenance measures can be taken, the accident risk is reduced, and the safety of a power system is improved.

Description

Technical Field

[0001] The embodiments of the present invention relate to the technical field of power equipment detection, and in particular to an insulator defect detection method. Background Art

[0002] Insulators are installed on transmission lines. In addition to bearing the impact of internal overvoltage, operating overvoltage, electrical overvoltage and long-term rated operating voltage, they also need to withstand the erosion of severe weather conditions such as strong winds, haze, ice, snow, and heavy rain. These severe weather conditions may cause insulators to have faults such as reduced insulation resistance, insulation cracking, and even breakdown. Over time, their insulation performance or mechanical properties will gradually decline, resulting in zero-value or low-value insulators, leading to the degradation of insulators and posing a potential threat to the reliability of power supply. In high-voltage transmission systems, if there are zero-value or low-value insulators in the insulator string, it is equivalent to short-circuiting part of the insulation of the insulator string, and the overall creepage distance will be reduced accordingly, thereby greatly increasing the probability of flashover. When flashover occurs, the arc may pass through the inside of the zero-value insulator, causing the steel cap on the insulator to often explode or detach, causing serious accidents such as string drop and conductor falling. Therefore, it is of great significance to detect the state of the insulator.

[0003] There are many methods for testing insulators, including insulation resistance method, voltage distribution method, pulse current method, ultrasonic testing method, ultraviolet imaging method and infrared thermal imaging method, etc. However, these methods have certain limitations in detection accuracy and cannot meet actual needs. Summary of the invention

[0004] The present invention provides an insulator defect detection method to solve the problem that the existing detection method has certain limitations in detection accuracy and cannot meet actual needs.

[0005] According to one aspect of the present invention, an embodiment of the present invention provides an insulator defect detection method, the method comprising:

[0006] Acquire first pre-detection data and second pre-detection data of the insulator; wherein the first pre-detection data and the second pre-detection data are pre-detection data corresponding to different types of pre-detection methods;

[0007] Encoding the first pre-detection data and the second pre-detection data to form a first feature vector and a second feature vector respectively;

[0008] Performing vector coupling on the first eigenvector and the second eigenvector to form a total vector;

[0009] The defect state of the insulator is determined based on the total vector.

[0010] Optionally, the first eigenvector and the second eigenvector are vector coupled to form a total vector, including: determining a first weight matrix and a second weight matrix according to the dimension of the first eigenvector, the dimension of the second eigenvector and a preset dimension; determining a first fusion vector according to the first weight matrix and the first eigenvector; determining a second fusion vector according to the second weight matrix and the second eigenvector; and fusing the first fusion vector and the second fusion vector to form the total vector.

[0011] Optionally, the first weight matrix and the second weight matrix are determined according to the dimension of the first eigenvector, the dimension of the second eigenvector and the preset dimension, including: determining the number of the first weight matrix and the second weight matrix according to the dimension of the first eigenvector and the dimension of the second eigenvector; determining the number of rows of the first weight matrix according to the dimension of the first eigenvector; determining the number of rows of the second weight matrix according to the dimension of the second eigenvector; determining the number of columns of the first weight matrix and the second weight matrix according to the preset dimension; using a first preset value as an element of the first weight matrix to form the first weight matrix; wherein different first weight matrices correspond to different first preset values; using a second preset value as an element of the second weight matrix to form the second weight matrix; wherein different second weight matrices correspond to different second preset values.

[0012] Optionally, determining the number of the first weight matrix and the second weight matrix according to the dimension of the first eigenvector and the dimension of the second eigenvector includes: taking the quotient of the dimension of the first eigenvector and the dimension of the second eigenvector as the number of the first weight matrix and the second weight matrix.

[0013] Optionally, determining a first fusion vector based on the first weight matrix and the first eigenvector includes: multiplying the first eigenvector with at least one of the first weight matrices to determine at least one first temporary vector; and adding elements of at least one of the first temporary vectors to determine a first fusion vector.

[0014] Optionally, determining a second fusion vector based on the second weight matrix and the second eigenvector includes: multiplying the second eigenvector with at least one of the second weight matrices to determine at least one second temporary vector; and adding elements of at least one of the second temporary vectors to determine a second fusion vector.

[0015] Optionally, the first fusion vector and the second fusion vector are fused to form the total vector, including: vector splicing the first fusion vector and the second fusion vector to determine the splicing vector; setting weights for the splicing vector in a fully connected layer of the encoder to form the total vector; wherein the weight is the product of the first preset value and the second preset value.

[0016] Optionally, fusing the first fused vector and the second fused vector to form the total vector includes: performing a vector inner product on the first fused vector and the second fused vector to form the total vector.

[0017] Optionally, determining the defect state of the insulator according to the total vector includes: comparing and matching the total vector with a defect type database to determine the defect state of the insulator.

[0018] Optionally, the first pre-detection data is ultrasonic detection data, and the second pre-detection data is infrared detection data.

[0019] The technical solution of this embodiment is to obtain the first pre-detection data and the second pre-detection data of the insulator; wherein the first pre-detection data and the second pre-detection data are pre-detection data corresponding to different types of pre-detection methods. Through two different types of detection methods, a more comprehensive and global detection of the insulator can be achieved. The first pre-detection data and the second pre-detection data are encoded, and the encoding can convert data of different formats or scales into a unified format to form a first eigenvector and a second eigenvector respectively. The first eigenvector and the second eigenvector are vector-coupled, and the vector coupling can fuse the characteristic information obtained by different methods to form a total vector, so that the total vector has the defect characteristics of the insulator represented by the first pre-detection data and the second pre-detection data at the same time, so as to more comprehensively represent the defect state of the insulator. The defect state of the insulator is determined according to the total vector, which improves the efficiency of data processing and the accuracy of detection. The technical solution of the embodiment of the present invention improves the detection accuracy of the insulator by combining the two detection methods, can timely discover the risk of faults, take corresponding maintenance measures, reduce the risk of accidents, and improve the safety of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 A flow chart of an insulator defect detection method provided by an embodiment of the present invention;

[0021] Figure 2 A flow chart of another insulator defect detection method provided by an embodiment of the present invention;

[0022] Figure 3 A schematic structural diagram of an insulator defect detection device provided by an embodiment of the present invention;

[0023] Figure 4 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0024] The present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the present invention, rather than to limit the present invention. It should also be noted that, for ease of description, only parts related to the present invention, rather than all structures, are shown in the accompanying drawings.

[0025] Figure 1 The flowchart of an insulator defect detection method provided by an embodiment of the present invention is applicable to the insulator defect state detection in a power grid. The method can be executed by an insulator defect detection device, such as Figure 1 As shown, the method specifically comprises the following steps:

[0026] S101. Acquire first pre-detection data and second pre-detection data of an insulator; wherein the first pre-detection data and the second pre-detection data are pre-detection data corresponding to different types of pre-detection methods.

[0027] Among them, the first pre-detection data and the second pre-detection data are pre-detection data obtained by different types of pre-detection methods. They can be obtained by different types of sensors, and different types of sensors are installed on the pole tower with insulators. The sensors upload data to the background every 10 minutes, and the background combines the data for analysis. Different types of pre-detection methods can be any two of the insulation resistance method, voltage distribution method, pulse current method, ultrasonic detection method, ultraviolet imaging method and infrared thermal imaging method. Exemplarily, the first pre-detection data and the second pre-detection data can be pre-detection data obtained by ultrasonic detection method and infrared thermal imaging method, and the sensors can be ultrasonic and infrared sensors.

[0028] Specifically, the first pre-detection data acquisition process may be to acquire ultrasonic data through an ultrasonic sound beam, place the probe at the detection position of the insulator, and adjust the angle and pressure of the probe as needed. Start the ultrasonic flaw detector and emit ultrasonic pulses. The ultrasonic pulse propagates inside the insulator and generates a reflected wave when encountering a defect. The probe receives the ultrasonic signal reflected by the defect. The flaw detector converts the received signal into an electrical signal and presents it as a waveform on the display screen as the first pre-detection data. The second pre-detection data acquisition process may be to scan the insulator using an infrared thermal imager and record its temperature distribution. Under normal circumstances, the temperature of the insulator should be uniform. If an abnormal increase in local temperature is found, this may indicate the presence of partial discharge or other faults. In an infrared image, abnormal hot spots are usually displayed as brighter areas than the surrounding areas. These hot spots may be caused by partial discharge, material defects, poor contact or other reasons. The detected temperature data and infrared images are recorded as the second pre-detection data.

[0029] S102 : Encode the first pre-detection data and the second pre-detection data to form a first feature vector and a second feature vector respectively.

[0030] Among them, an encoder can be used to encode the first pre-detection data and the second pre-detection data to extract features from the data. Encoding is to convert high-dimensional and complex data into low-dimensional and more abstract feature representations. Before feature extraction, the first pre-detection data and the second pre-detection data usually need to be preprocessed. This may include normalization, standardization, denoising, filling missing values, downsampling and other operations to ensure that the data is suitable for input into the encoder. The encoder can be a simple linear model or a complex deep neural network. In deep learning, the encoder is usually composed of a multi-layer neural network, including convolutional layers, pooling layers, fully connected layers, recurrent layers, etc. Each layer is responsible for extracting features at different levels from the input data. The encoder learns the features of the data through a forward propagation process. At each layer, the model applies a series of transformations and activation functions to convert the input data into a new feature representation. For example, a convolutional layer can extract local features, a pooling layer can reduce the spatial dimension of the features, and a fully connected layer can map the features to a higher-dimensional space.

[0031] Specifically, one of the purposes of the encoder is dimensionality reduction, that is, to represent the data in a more compact form. This can be achieved by using fewer neurons, applying specific activation functions, or using specific network layers when designing the network structure. Dimensionality reduction is a data analysis technique that simplifies a data set by reducing the number of features of the data while retaining the important information of the data as much as possible. Dimensionality reduction helps to remove redundant information and retain the most important features. After the encoder training is completed, the features of the new data can be extracted by dimensionality reduction. After the input data passes through the encoder, the low-dimensional representation of the output is the feature vector. Exemplarily, the first feature vector after the first pre-detection data is encoded can be a 1*m vector, and the second feature vector after the second pre-detection data is encoded can be a 1*n vector.

[0032] S103 , performing vector coupling on the first eigenvector and the second eigenvector to form a total vector.

[0033] Among them, vector coupling generally refers to combining two eigenvectors in some way, which can be in the form of vector concatenation or summation, and combining the first eigenvector and the second eigenvector to obtain a total vector. The total vector has the insulator defect characteristics represented by the first pre-detection data and the second pre-detection data at the same time, so as to more comprehensively represent the defect state of the insulator.

[0034] S104. Determine the defect state of the insulator according to the total vector.

[0035] Specifically, the defect state of the insulator is compared and matched with the defect type database according to the total vector to obtain the defect state type of the insulator. Among them, the defect type database refers to a database system used to store and manage system defect information. It contains the attributes of the defect, such as defect description, defect type, severity, etc. The total vector has the insulator defect characteristics represented by the first pre-detection data and the second pre-detection data. The defect state of the insulator is determined by the total vector, which avoids the tediousness and time cost of manual inspection. It can detect the defect state of the insulator more comprehensively, and the monitoring data can be automatically recorded and analyzed to provide a basis for fault diagnosis and maintenance decisions. Potential fault risks can be discovered in time, and corresponding maintenance measures can be taken to reduce accident risks and improve the safety of the power system.

[0036] The technical solution of this embodiment is to obtain the first pre-detection data and the second pre-detection data of the insulator; wherein the first pre-detection data and the second pre-detection data are pre-detection data corresponding to different types of pre-detection methods. Through two different types of detection methods, a more comprehensive and global detection of the insulator can be achieved. The first pre-detection data and the second pre-detection data are encoded, and the encoding can convert data of different formats or scales into a unified format to form a first eigenvector and a second eigenvector respectively. The first eigenvector and the second eigenvector are vector-coupled, and the vector coupling can fuse the characteristic information obtained by different methods to form a total vector, so that the total vector has the defect characteristics of the insulator represented by the first pre-detection data and the second pre-detection data at the same time, so as to more comprehensively represent the defect state of the insulator. The defect state of the insulator is determined according to the total vector, which improves the efficiency of data processing and the accuracy of detection. The technical solution of the embodiment of the present invention improves the detection accuracy of the insulator by combining the two detection methods, can timely discover the risk of faults, take corresponding maintenance measures, reduce the risk of accidents, and improve the safety of the power system.

[0037] Figure 2 This is a flow chart of another insulator defect detection method provided by an embodiment of the present invention. This embodiment further refines the insulator defect detection method based on the above embodiments. Figure 2 As shown, the method includes:

[0038] S201. Acquire first pre-detection data and second pre-detection data of an insulator; wherein the first pre-detection data and the second pre-detection data are pre-detection data corresponding to different types of pre-detection methods.

[0039] S202 : Encode the first pre-detection data and the second pre-detection data to form a first feature vector and a second feature vector respectively.

[0040] S2031. Determine a first weight matrix and a second weight matrix according to a dimension of the first eigenvector, a dimension of the second eigenvector, and a preset dimension.

[0041] The dimension of the eigenvector is determined by the dimension of the matrix to which it corresponds, and the preset dimension usually refers to the dimension expected by the insulator in a specific case. For example, if the first eigenvector is a 1*m vector and the second eigenvector is a 1*n vector, the dimension of the first eigenvector is m, the dimension of the second eigenvector is n, and the preset dimension can be h.

[0042] In some optional embodiments of the present invention, determining the first weight matrix and the second weight matrix according to the dimension of the first eigenvector, the dimension of the second eigenvector and the preset dimension includes:

[0043] The number of the first weight matrix and the second weight matrix is ​​determined according to the dimension of the first eigenvector and the dimension of the second eigenvector.

[0044] Optionally, determining the number of first weight matrices and second weight matrices according to the dimension of the first eigenvector and the dimension of the second eigenvector includes: taking the quotient of the dimension of the first eigenvector and the dimension of the second eigenvector as the number of first weight matrices and second weight matrices.

[0045] Among them, the first eigenvector is a 1*m vector, and the second eigenvector is a 1*n vector, then the dimension of the first eigenvector is m, the dimension of the second eigenvector is n, and the number of the first weight matrix and the second weight matrix is ​​r=m / n.

[0046] The number of rows of the first weight matrix is ​​determined according to the dimension of the first eigenvector.

[0047] Among them, the dimension of the first eigenvector is m, and the number of rows of the first weight matrix is ​​m.

[0048] The number of rows of the second weight matrix is ​​determined according to the dimension of the second eigenvector.

[0049] Among them, the dimension of the second eigenvector is n, and the number of rows of the second weight matrix is ​​n.

[0050] Determine the number of columns of the first weight matrix and the second weight matrix according to a preset dimension;

[0051] Among them, the preset dimension is h, and the number of columns of the first weight matrix and the second weight matrix is ​​h.

[0052] Using the first preset value as an element of a first weight matrix to form a first weight matrix; wherein different first weight matrices correspond to different first preset values;

[0053] The first weight matrix has m rows and h columns, and is an m*h matrix. The first weight matrix may be r m*h matrices. Different first weight matrices correspond to different first preset values.

[0054] The second preset values ​​are used as elements of the second weight matrix to form a second weight matrix; wherein different second weight matrices correspond to different second preset values.

[0055] The number of rows of the second weight matrix is ​​n, and the number of columns is h, then the second weight matrix is ​​an n*h matrix, and the second weight matrix may be r n*h matrices. Different second weight matrices correspond to different second preset values.

[0056] S2032 determines a first fusion vector according to the first weight matrix and the first eigenvector.

[0057] The first eigenvector and the first weight matrix are processed to obtain a first fusion vector, and the processing may include at least one of addition, subtraction, multiplication or division operations.

[0058] Optionally, determining the first fusion vector according to the first weight matrix and the first eigenvector includes:

[0059] The first eigenvector is multiplied by at least one first weight matrix to determine at least one first temporary vector; and elements of the at least one first temporary vector are added to determine a first fusion vector.

[0060] The first eigenvector is multiplied by the first weight matrix to obtain the first temporary vector. The elements of the first temporary vector are added to obtain the first fusion vector. Exemplarily, the first eigenvector is a 1*m vector, which is multiplied by r first weight matrices m*h to obtain r 1*h first temporary vectors. 11 、m 12 ……m 1h The corresponding m r1 、m r2 ……m rh Add each element to get a first fused vector of 1*h. When r is 1, the first temporary vector is the first fused vector.

[0061] S2033. Determine a second fusion vector according to the second weight matrix and the second eigenvector.

[0062] The second eigenvector and the second weight matrix are processed to obtain a second fusion vector, and the processing may include at least one of addition, subtraction, multiplication or division operations.

[0063] Optionally, determining the second fusion vector according to the second weight matrix and the second eigenvector includes:

[0064] The second eigenvector is multiplied by at least one second weight matrix to determine at least one second temporary vector; and elements of the at least one second temporary vector are added to determine a second fusion vector.

[0065] The second eigenvector is multiplied by the second weight matrix to obtain a second temporary vector. The elements of the second temporary vector are added to obtain a second fusion vector. Exemplarily, the second eigenvector is a 1*n vector, which is multiplied by r second weight matrices n*h to obtain r 1*h second temporary vectors. 11 、n 12 ……n 1h And the corresponding n r1 、n r2 ……n rhAdd each element to get a second fused vector of 1*h. When r is 1, the second temporary vector is the second fused vector.

[0066] S2034: Fuse the first fused vector and the second fused vector to form a total vector.

[0067] Fusion is the process of combining feature vectors from different sources or different feature extraction methods to obtain a richer and more comprehensive feature representation. Fusion can include operations such as vector concatenation and vector inner product. The total vector is obtained by fusing the first fused vector and the second fused vector.

[0068] In some optional embodiments of the present invention, the first fused vector and the second fused vector are fused to form a total vector, including:

[0069] Performing vector splicing on the first fusion vector and the second fusion vector to determine a splicing vector;

[0070] The concatenated vectors are weighted in the fully connected layer of the encoder to form a total vector, wherein the weight is the product of the first preset value and the second preset value.

[0071] Among them, vector splicing involves connecting multiple vectors to form a longer vector. This method is often used to merge features from different feature sets, or to combine the outputs of multiple models for further analysis or prediction. The first fused vector and the second fused vector are two vectors obtained from the processing of pre-detection data corresponding to different types of pre-detection methods. To ensure compatibility, before splicing, ensure that the first fused vector and the second fused vector are compatible in the splicing dimension. For example, if the vectors are spliced ​​in the horizontal direction, the lengths of the first fused vector and the second fused vector must be the same. If the vector lengths are different, padding or truncation operations are required. For example, if the first fused vector is 1*p and the second fused vector is 1*q, where q is greater than p, the first fused vector can be padded to 1*q and then vector spliced ​​with the second fused vector. The second fused vector can also be truncated to 1*p and then vector spliced ​​with the first fused vector.

[0072] Specifically, in an embodiment of the present invention, the vector splicing of the first fused vector and the second fused vector can be horizontal or vertical. If horizontal splicing is adopted, the first fused vector and the second fused vector are placed on the same horizontal line to form a new, longer vector, which is similar to the horizontal splicing of a matrix. If vertical splicing is adopted, the first fused vector and the second fused vector are stacked to form a new, higher-dimensional vector, which is similar to the vertical splicing of a matrix. Exemplarily, the first fused vector of 1*h is horizontally spliced ​​with the second fused vector of 1*h to obtain a splicing vector of 1*2h. The first fused vector of 1*h is vertically spliced ​​with the second fused vector of 1*h to obtain a vector of 2*h. If horizontal splicing is adopted, the splicing vector is a vector of 1*2h, and the splicing vector is weighted in the fully connected layer of the encoder, and the weight of u×v is multiplied in the fully connected layer to obtain a total vector 1*h. u is a first preset value, and v is a second preset value.

[0073] In some optional embodiments of the present invention, the first fused vector and the second fused vector are fused to form a total vector, including: performing vector inner product on the first fused vector and the second fused vector to form a total vector.

[0074] The vector inner product, also called the dot product or scalar product, is an operation between two vectors. It multiplies two vectors to get a scalar. The vector inner product of the first fused vector and the second fused vector is performed to get the total vector.

[0075] In some optional embodiments of the present invention, determining the defect state of the insulator according to the total vector includes:

[0076] The total vector is compared and matched with the defect type database to determine the defect status of the insulator.

[0077] Among them. The defect type database is a database system used to store and manage system defect information. It contains the attributes of the defect, such as defect description, defect type, severity, etc. The determination of the defect type database includes: constructing insulators with different defect types; controlling the insulators to work at the set operating voltage, and measuring the insulators through two different pre-detection methods to obtain the characteristic data of insulators with different defect types; based on the acquired characteristic data, obtaining the characteristic frequency corresponding to each defect type, and storing them accordingly to form a defect type database.

[0078] S204. Determine the defect state of the insulator according to the total vector.

[0079] The technical solution of this embodiment determines the first weight matrix and the second weight matrix according to the dimension of the first eigenvector, the dimension of the second eigenvector and the preset dimension; specific problems can be solved. The first fusion vector is determined according to the first weight matrix and the first eigenvector, and the second fusion vector is determined according to the second weight matrix and the second eigenvector. The original feature space can be mapped to the new feature space, which helps to effectively extract information. The first fusion vector and the second fusion vector are fused to form a total vector. Vector fusion can reduce the amount of data that needs to be processed, thereby improving detection accuracy. The technical solution of the embodiment of the present invention improves the accuracy and reliability of the detection results and the reliability and safety of the power system by combining two detection technologies and processing and fusing the detection data.

[0080] In an optional embodiment of the present invention, after comparing and matching the total vector with the defect type database to determine the defect state of the insulator, the method further includes:

[0081] Obtain the actual defect state corresponding to the insulator to be detected; calculate the loss function according to the above-determined insulator defect state and the actual defect state; and update the parameters of the defect type database and / or the generated weight matrix using the loss function through a preset parameter adjustment strategy.

[0082] Among them, the loss function is a function used in machine learning to measure the difference between the model prediction value and the actual value. The loss function can be any loss function. Through the preset parameter adjustment strategy, the loss function of each data sample is used to update the parameters of the defect type database and / or the generated weight matrix, so as to make the total vector generation method or the model recognition effect better.

[0083] In some optional embodiments of the present invention, the first pre-detection data is ultrasonic detection data, and the second pre-detection data is infrared detection data.

[0084] Among them, ultrasonic detection data can be used to detect defects and deterioration inside insulators. By sending ultrasonic signals and receiving echoes, air bubbles, cracks, foreign objects and other problems inside insulators can be detected. Ultrasonic detection technology can provide a comprehensive understanding of the internal structure of insulators, help to find potential fault points early, and prevent insulator failure. Infrared detection data can be used to detect the surface temperature distribution of insulators, thereby evaluating the working status of insulators. Through infrared thermal imaging technology, the heat distribution on the surface of the insulator can be observed in real time to detect whether there is abnormal temperature rise and whether there is local overheating. Including both ultrasonic detection data and infrared detection data, the surface and internal conditions of the insulator can be detected at the same time. Ultrasonic waves can propagate into the inside of the insulator and detect potential defects, such as material voids, cracks, foreign objects, etc. Infrared rays can detect the external temperature distribution of the insulator, such as local overheating.

[0085] Figure 3 The present invention provides a schematic diagram of the structure of an insulator defect detection device. Figure 3 As shown, the device comprises:

[0086] The data acquisition module 301 is used to acquire first pre-detection data and second pre-detection data of the insulator; wherein the first pre-detection data and the second pre-detection data are pre-detection data corresponding to different types of pre-detection methods.

[0087] The data encoding module 302 is used to encode the first pre-detection data and the second pre-detection data to form a first feature vector and a second feature vector respectively.

[0088] The vector coupling module 303 is used to perform vector coupling on the first eigenvector and the second eigenvector to form a total vector.

[0089] The state determination module 304 is used to determine the defect state of the insulator according to the total vector.

[0090] Optionally, the vector coupling module 303 specifically includes:

[0091] The weight matrix determination module is used to determine the first weight matrix and the second weight matrix according to the dimension of the first eigenvector, the dimension of the second eigenvector and the preset dimension.

[0092] The first fusion vector determination module is used to determine the first fusion vector according to the first weight matrix and the first eigenvector.

[0093] The second fusion vector determination module is used to determine the second fusion vector according to the second weight matrix and the second eigenvector.

[0094] The total vector determination module is used to fuse the first fused vector and the second fused vector to form a total vector.

[0095] Optionally, the weight matrix determination module specifically includes:

[0096] The weight matrix quantity determination module is used to determine the quantity of the first weight matrix and the second weight matrix according to the dimension of the first eigenvector and the dimension of the second eigenvector.

[0097] The first row number determination module is used to determine the number of rows of the first weight matrix according to the dimension of the first eigenvector.

[0098] The second row number determination module is used to determine the number of rows of the second weight matrix according to the dimension of the second eigenvector.

[0099] The column number determination module is used to determine the number of columns of the first weight matrix and the second weight matrix according to a preset dimension.

[0100] The first weight matrix determination module is used to use the first preset value as an element of the first weight matrix to form a first weight matrix; wherein different first weight matrices correspond to different first preset values.

[0101] The second weight matrix determination module uses the second preset value as an element of the second weight matrix to form the second weight matrix; wherein different second weight matrices correspond to different second preset values.

[0102] Optionally, the weight matrix quantity determination module is specifically used for:

[0103] The quotient of the dimension of the first eigenvector and the dimension of the second eigenvector is taken as the number of the first weight matrix and the second weight matrix.

[0104] Optionally, the first fusion vector determination module is specifically used for:

[0105] The first eigenvector is multiplied by at least one first weight matrix to determine at least one first temporary vector; and elements of the at least one first temporary vector are added to determine a first fusion vector.

[0106] Optionally, the second fusion vector determination module is specifically used for:

[0107] The second eigenvector is multiplied by at least one second weight matrix to determine at least one second temporary vector; and elements of the at least one second temporary vector are added to determine a second fusion vector.

[0108] Optionally, the total vector determination module is specifically used for:

[0109] The first fusion vector and the second fusion vector are vector-concatenated to determine a concatenated vector; and the concatenated vector is weighted in a fully connected layer of the encoder to form a total vector; wherein the weight is the product of a first preset value and a second preset value.

[0110] Optionally, the total vector determination module is further used for:

[0111] Perform the vector inner product of the first fused vector and the second fused vector to form a total vector.

[0112] Optionally, the state determination module is specifically used for:

[0113] The total vector is compared and matched with the defect type database to determine the defect status of the insulator.

[0114] Figure 4 It is a structural schematic diagram of an electronic device provided by an embodiment of the present invention. Figure 4 A block diagram of an exemplary electronic device 12 suitable for use in implementing embodiments of the present invention is shown. Figure 4 The electronic device 12 shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0115] like Figure 3 As shown, the electronic device 12 is in the form of a general purpose computing device. The components of the electronic device 12 may include, but are not limited to: one or more processors 16, a memory 28, and a bus 18 that connects various system components (including the memory 28 and the processor 16).

[0116] Bus 18 represents one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processor, or a local bus using any of a variety of bus architectures. For example, these architectures include, but are not limited to, Industry Standard Architecture (ISA) bus, Micro Channel Architecture (MCA) bus, Enhanced ISA bus, Video Electronics Standards Association (VESA) local bus, and Peripheral Component Interconnect (PCI) bus.

[0117] The electronic device 12 typically includes a variety of computer system readable media. These media can be any available media that can be accessed by the electronic device 12, including volatile and non-volatile media, removable and non-removable media.

[0118] The memory 28 may include computer system readable media in the form of volatile memory, such as random access memory (RAM) 30 and / or cache memory 32. The electronic device 12 may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, the storage system 34 may be used to read and write non-removable, non-volatile magnetic media ( Figure 3 not shown, usually called a "hard drive"). Although Figure 3 Not shown in the figure, a disk drive for reading and writing a removable non-volatile disk (e.g., a "floppy disk"), and an optical disk drive for reading and writing a removable non-volatile optical disk (e.g., a read-only optical disk (Compact Disc-Read Only Memory, CD-ROM), a digital video disk (Digital Video Disc-Read Only Memory, DVD-ROM) or other optical media) may be provided. In these cases, each drive may be connected to the bus 18 via one or more data medium interfaces. The memory 28 may include at least one program product having a set (e.g., at least one) of program modules that are configured to perform the functions of the various embodiments of the present invention.

[0119] A program / utility 40 having a set (at least one) of program modules 42 may be stored, for example, in the memory 28, such program modules 42 including, but not limited to, an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment. The program modules 42 generally perform the functions and / or methods of the embodiments described herein.

[0120] The electronic device 12 may also communicate with one or more external devices 14 (e.g., keyboards, pointing devices, displays 24, etc.), one or more devices that enable a user to interact with the electronic device 12, and / or any device that enables the electronic device 12 to communicate with one or more other computing devices (e.g., a network card, a modem, etc.). Such communication may be performed through an input / output (I / O) interface 22. Furthermore, the electronic device 12 may also communicate with one or more networks (e.g., a local area network (LAN), a wide area network (WAN) and / or a public network, such as the Internet) through a network adapter 20. As shown, the network adapter 20 communicates with other modules of the electronic device 12 through the bus 18. It should be understood that although Figure 3Not shown, other hardware and / or software modules may be used in conjunction with the electronic device 12, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, (Redundant Arrays of Independent Disks, RAID) systems, tape drives, and data backup storage systems.

[0121] The processor 16 executes various functional applications and data processing by running the program stored in the memory 28 to implement the insulator defect detection method provided by the embodiment of the present invention: obtaining first pre-detection data and second pre-detection data of the insulator; wherein the first pre-detection data and the second pre-detection data are pre-detection data corresponding to different types of pre-detection methods; encoding the first pre-detection data and the second pre-detection data to form a first eigenvector and a second eigenvector respectively; performing vector coupling on the first eigenvector and the second eigenvector to form a total vector; and determining the defect state of the insulator according to the total vector.

[0122] An embodiment of the present invention further provides a computer storage medium storing a computer program, wherein the computer program, when executed by a computer processor, is used to execute the insulator defect detection method described in any of the above embodiments of the present invention: obtaining first pre-detection data and second pre-detection data of the insulator; wherein the first pre-detection data and the second pre-detection data are pre-detection data corresponding to different types of pre-detection methods; encoding the first pre-detection data and the second pre-detection data to form a first eigenvector and a second eigenvector, respectively; performing vector coupling on the first eigenvector and the second eigenvector to form a total vector; and determining the defect state of the insulator according to the total vector.

[0123] The computer storage medium of the embodiment of the present invention can adopt any combination of one or more computer-readable media. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium can be, for example, but not limited to, a system, device or device of electricity, magnetism, light, electromagnetic, infrared, or semiconductor, or any combination of the above. More specific examples (non-exhaustive list) of computer-readable storage media include: an electrical connection with one or more wires, 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 above. In this document, a computer-readable storage medium can be any tangible medium containing or storing a program, which can be used by an instruction execution system, a device or a device or used in combination with it.

[0124] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, which carry computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0125] The program code contained on the computer-readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wire, optical cable, radio frequency (RF), etc., or any suitable combination of the foregoing.

[0126] Computer program code for performing the operations of the present invention may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0127] Note that the above are only preferred embodiments of the present invention and the technical principles used. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein, and that various obvious changes, readjustments and substitutions can be made by those skilled in the art without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in more detail through the above embodiments, the present invention is not limited to the above embodiments, and may include more other equivalent embodiments without departing from the concept of the present invention, and the scope of the present invention is determined by the scope of the appended claims.

Claims

1. A method for detecting insulator defects, characterized in that: include: Acquire first pre-detection data and second pre-detection data of the insulator; wherein the first pre-detection data and the second pre-detection data are pre-detection data corresponding to different types of pre-detection methods; Encoding the first pre-detection data and the second pre-detection data to form a first feature vector and a second feature vector respectively; Performing vector coupling on the first eigenvector and the second eigenvector to form a total vector; The defect state of the insulator is determined according to the total vector.

2. The insulator defect detection method according to claim 1, characterized in that: Performing vector coupling on the first eigenvector and the second eigenvector to form a total vector includes: Determine a first weight matrix and a second weight matrix according to the dimension of the first eigenvector, the dimension of the second eigenvector and a preset dimension; Determine a first fusion vector according to the first weight matrix and the first eigenvector; Determine a second fusion vector according to the second weight matrix and the second eigenvector; The first fused vector and the second fused vector are fused to form the total vector.

3. The insulator defect detection method according to claim 2, characterized in that: Determining a first weight matrix and a second weight matrix according to a dimension of the first eigenvector, a dimension of the second eigenvector, and a preset dimension includes: Determining the number of the first weight matrix and the second weight matrix according to the dimension of the first eigenvector and the dimension of the second eigenvector; Determining the number of rows of the first weight matrix according to the dimension of the first eigenvector; Determining the number of rows of the second weight matrix according to the dimension of the second eigenvector; Determining the number of columns of the first weight matrix and the second weight matrix according to the preset dimension; Using the first preset value as an element of the first weight matrix to form the first weight matrix; wherein different first weight matrices correspond to different first preset values; The second preset value is used as an element of the second weight matrix to form the second weight matrix; wherein different second weight matrices correspond to different second preset values.

4. The insulator defect detection method according to claim 3, characterized in that: Determining the number of the first weight matrix and the second weight matrix according to the dimension of the first eigenvector and the dimension of the second eigenvector includes: The quotient of the dimension of the first eigenvector and the dimension of the second eigenvector is taken as the number of the first weight matrix and the number of the second weight matrix.

5. The insulator defect detection method according to claim 3, characterized in that: Determining a first fusion vector according to the first weight matrix and the first eigenvector includes: multiplying the first eigenvector by at least one of the first weight matrices to determine at least one first temporary vector; Elements of at least one of the first temporary vectors are added together to determine a first fused vector.

6. The insulator defect detection method according to any one of claims 3 to 5, characterized in that: Determining a second fusion vector according to the second weight matrix and the second eigenvector includes: multiplying the second eigenvector by at least one of the second weight matrices to determine at least one second temporary vector; The elements of at least one of the second temporary vectors are added together to determine a second fused vector.

7. The insulator defect detection method according to claim 6, characterized in that: The first fused vector and the second fused vector are fused to form the total vector, comprising: Performing vector splicing on the first fusion vector and the second fusion vector to determine a splicing vector; The concatenated vector is weighted at a fully connected layer of the encoder to form the total vector; wherein the weight is the product of the first preset value and the second preset value.

8. The insulator defect detection method according to claim 6, characterized in that: The first fused vector and the second fused vector are fused to form the total vector, comprising: Perform a vector inner product on the first fused vector and the second fused vector to form the total vector.

9. The insulator defect detection method according to claim 1, characterized in that: Determining the defect state of the insulator according to the total vector includes: The total vector is compared and matched with a defect type database to determine the defect state of the insulator.

10. The insulator defect detection method according to claim 1, characterized in that: The first pre-detection data is ultrasonic detection data, and the second pre-detection data is infrared detection data.

Images