High-precision power transmission hidden danger and defect identification method and system

Abstract

The invention discloses a high-precision power transmission hidden danger and defect identification method and system, and relates to the field of power transmission hidden danger and defect identification, and the method comprises the steps: obtaining the abnormal condition pictures of related equipment and devices in the construction process of a power transmission line; picture pixel information is extracted, the picture is subjected to matrix processing, and a spatial domain signal of the picture is converted into a frequency domain signal; based on the picture ambiguity two-dimensional ambiguity index, identifying whether a picture sharpness requirement is met or not when picture acquisition is carried out on the power transmission line and the related equipment; denoising and optimizing the picture data according to the filtering array expression, and determining a picture data input feature matrix; and according to the picture data input feature matrix, identifying whether the power transmission line and related equipment have hidden dangers and defects based on the neural network model. The method has the advantages that the collected pictures are subjected to noise reduction and optimization, so that the accuracy of power transmission hidden danger and defect identification is effectively improved, and an abnormal alarm is given in time.

Description

Technical Field

[0001] The present invention relates to the field of power transmission hidden danger and defect identification, and in particular to a high-precision power transmission hidden danger and defect identification method and system. Background Art

[0002] Nowadays, electricity has become closely related to everyone, from the country to the people. The stability of electricity is an indispensable guarantee for economic development, people's livelihood, engineering construction, national security, etc., and the guarantee of electricity stability is to timely identify and deal with hidden dangers and defects in the power transportation process. With the rise of drones, robots and visual recognition technologies, drones, robots and visual recognition technologies are applied to the identification of hidden dangers and defects in high-altitude power transmission, replacing traditional manual inspections, which greatly improves the efficiency and resource costs of identifying hidden dangers and defects in high-altitude power transmission. Therefore, the use of drones, robots and other technologies to integrate detection equipment and detect power transmission hidden dangers and defects through visual recognition has received a lot of attention. It will become crucial to improve the recognition accuracy of detection equipment and reduce the noise impact generated by drones and robots during movement.

[0003] During the image acquisition process, due to the influence of the complex external environment or the influence of the drone and robot themselves during the movement process, certain non-stationary noise errors will be generated, which will lead to errors in the final judgment. The existing technology for filtering images mainly focuses on Gaussian filtering and Fourier filtering, but this type of filtering algorithm tends to have a filtering effect on a stable state. There will be errors in the filtering effect on non-stationary states that may exist during the movement process, resulting in unclear collected image information, which in turn causes inaccurate final judgment results, easily resulting in missed detections and abnormal alarms, which is not conducive to the long-term development of power transmission hidden dangers and defect identification. Summary of the invention

[0004] In order to solve the above technical problems, a high-precision method and system for identifying power transmission hazards and defects are provided. The technical solution solves the problem proposed in the above background technology that this type of filtering algorithm tends to have a filtering effect on a stable state. Errors will occur in the filtering effect on a non-stable state that may exist during the movement process, resulting in unclear collected image information, which in turn causes the final judgment result to be inaccurate, easily resulting in missed detections and abnormal alarms, and is not conducive to the long-term development of power transmission hazard and defect identification.

[0005] In order to achieve the above purpose, the technical solution adopted by the present invention is:

[0006] A high-precision power transmission hidden danger and defect identification method, characterized by comprising:

[0007] Based on big data, obtain abnormal situation pictures of related equipment and devices during the installation of transmission lines;

[0008] Extract image pixel information, perform matrix processing on the image, and convert the image's spatial domain signal into a frequency domain signal;

[0009] Based on the two-dimensional fuzziness index of the image fuzziness, it is identified whether the image clarity requirements are met when collecting images of the transmission lines and related equipment. If so, the requirements are met. If not, the collection angle and accuracy of the detection equipment need to be adjusted according to the actual situation;

[0010] According to the filter array expression, the image data is denoised and optimized to determine the input feature matrix of the image data;

[0011] According to the feature matrix input by the image data, based on the neural network model, it is identified whether there are hidden dangers and defects in the transmission lines and related equipment.

[0012] Preferably, extracting the pixel information of the picture, performing matrix processing on the picture, and converting the spatial domain signal of the picture into a frequency domain signal specifically includes:

[0013] Get the image pixel information and grayscale the red, green, and blue components of each pixel;

[0014] Based on the Gaussian filtering algorithm, the gray value of the Gaussian filtered origin of the n×n window coordinates is calculated;

[0015] According to the gray value after Gaussian filtering, set the n×n gray value input matrix;

[0016] According to the Fourier transform formula, the spatial n×n grayscale value input matrix and the frequency domain n×n grayscale value input matrix are interchanged;

[0017] According to the n×n grayscale value input matrix in the frequency domain, based on the components of the image data in the frequency domain, the noise value in the stable state is filtered out;

[0018] The gray value expression after calculating the Gaussian filter of the n×n window coordinate origin is:

[0019]

[0020] Where f(x,y) is the grayscale value of the image that obeys the two-dimensional Gaussian function, g(i,j) is the grayscale value of the image after Gaussian filtering in the i-th row and j-th column, (x0,y0) is the coordinate origin in the n×n window, (x i ,y j ) is the coordinate information of the grayscale value of the image corresponding to the i-th row and j-th column, σ is the standard deviation, p(i,j) is the grayscale value probability of the i-th row and j-th column image, and n is the number of grayscale values;

[0021] The n×n grayscale value input matrix expression is:

[0022]

[0023] Where G(i,j) is the n×n gray value matrix of the image after Gaussian filtering;

[0024] The conversion formula of the spatial input matrix into the frequency domain input matrix is:

[0025]

[0026] Where F(u,v) is the gray value matrix in the frequency domain, n is the number of gray values, and G(i,j) is the n×n gray value matrix of the image after Gaussian filtering.

[0027] Preferably, the identification of whether the image clarity requirement is met when collecting images of the power transmission line and related equipment based on the two-dimensional fuzziness index of the image fuzziness specifically includes:

[0028] Obtain the standard grayscale value range of the same position images of the same transmission line and related equipment collected at different speeds and angles;

[0029] According to the standard gray value interval, set the two-dimensional blur threshold of the image blur;

[0030] Based on the gray value matrix in the frequency domain, determine the two-dimensional blur index of the image blur;

[0031] According to the comparison result of the two-dimensional blur index of the image blur and the two-dimensional blur threshold of the image blur, it is identified whether the image clarity requirements are met when collecting images of the transmission line and related equipment, and whether the two-dimensional blur index is less than the two-dimensional blur threshold. If so, the requirements are met. If not, the collection angle and accuracy of the detection equipment need to be adjusted according to the actual situation;

[0032] The two-dimensional fuzziness index expression of the image fuzziness is:

[0033]

[0034] Where S is the two-dimensional fuzzy index of the image fuzziness, β is the two-dimensional fuzzy deviation factor of the image fuzziness, is a constant term, N is the number of standard grayscale values ​​of the same position image collected at different speeds and angles for the same transmission line and related equipment, and F(u,v) is the two-dimensional grayscale value in the frequency domain.

[0035] Preferably, the denoising and optimizing the image data according to the filter array expression to determine the image data input feature matrix specifically includes:

[0036] According to the actual situation of the load device, install the array image acquisition equipment;

[0037] According to the two-dimensional wavelet transform formula, the collected images are denoised and optimized in a non-stationary state;

[0038] According to the judgment result of the two-dimensional blur index of the image blur, the credibility weight is set and the filter array expression is established;

[0039] According to the filter array expression, determine the image data input feature matrix;

[0040] The two-dimensional wavelet transform formula is:

[0041]

[0042] Where WT(α,τ) is the input matrix of wavelet transform for n×n grayscale value input matrix, α and τ are the variable parameters after wavelet transform; G(i,j) is the n×n grayscale value input matrix, (x,y) is the coordinate information of the corresponding image grayscale value;

[0043] The filter array expression is:

[0044]

[0045] Where Z is the grayscale value input matrix of the filter array, S is the two-dimensional fuzzy index matrix of the image fuzziness, m is the number of array image acquisition devices, and F(u,v) and WT(α,τ) are the Fourier filtering results and wavelet filtering results of the image data information collected at the same position, respectively.

[0046] Preferably, the step of inputting a feature matrix according to the image data and identifying whether there are hidden dangers and defects in the power transmission lines and related equipment based on a neural network model specifically includes:

[0047] According to the gray value input matrix formula of the filter array, the characteristic information of the abnormal situation pictures of the relevant equipment and devices in the process of erecting different transmission lines is extracted;

[0048] According to the characteristic information of the abnormal situation pictures, a characteristic information database of the abnormal situation pictures is established;

[0049] Based on the neural network model, the grayscale value input matrix of the filter array in the feature information library of the abnormal situation picture is trained and learned;

[0050] According to the trained neural network model, the newly input filter array grayscale value input matrix is ​​judged to identify whether there are hidden dangers and defects in the transmission line and related equipment. If so, it means that the line and equipment are operating normally. If not, it means that the line and equipment have hidden dangers and defects. By comparing the abnormal type of the feature information library of the abnormal situation picture, a problem warning is sent.

[0051] Furthermore, a high-precision power transmission hidden danger and defect identification system is proposed, which is used to implement the high-precision power transmission hidden danger and defect identification method as described above, including:

[0052] A data acquisition module, which is used to obtain abnormal situation pictures of related equipment and devices during the installation of power transmission lines based on big data;

[0053] A data processing module is used to extract pixel information of the image, perform matrix processing on the image, and convert the spatial domain signal of the image into a frequency domain signal; based on the two-dimensional fuzziness index of the image fuzziness, identify whether the image clarity requirements are met when collecting images of the transmission line and related equipment. If so, the requirements are met; if not, the collection angle and accuracy of the detection equipment need to be adjusted according to the actual situation; according to the filter array expression, the image data is denoised and optimized to determine the input feature matrix of the image data;

[0054] The anomaly detection module is used to input a feature matrix according to the image data and identify whether there are hidden dangers and defects in the transmission lines and related equipment based on the neural network model.

[0055] Preferably, the data processing module specifically includes:

[0056] A signal conversion unit, the signal conversion unit is used to extract the pixel information of the picture, perform matrix processing on the picture, and convert the spatial domain signal of the picture into a frequency domain signal;

[0057] A fuzzy index unit, which is used to identify whether the image clarity requirements are met when collecting images of power transmission lines and related equipment based on a two-dimensional fuzzy index of image fuzziness. If so, the requirements are met; if not, the collection angle and accuracy of the detection equipment need to be adjusted according to actual conditions;

[0058] An input matrix unit is used to denoise and optimize the image data according to a filter array expression to determine an input feature matrix of the image data.

[0059] Compared with the prior art, the present invention has the following beneficial effects:

[0060] The present invention proposes a high-precision power transmission hidden danger and defect identification method, which reduces the noise pollution caused by jitter, movement or external environmental factors by filtering and analyzing the abnormal situation pictures of related equipment and devices collected during the installation of power transmission lines. Among them, this scheme uses the Fourier transform formula to convert the two-dimensional spatial domain information of the image data into a frequency domain signal that is convenient for filtering processing. At the same time, the two-dimensional fuzzy index of the image fuzziness is set according to the transformed frequency domain signal to judge whether the current signal is stable and whether better image data information can be obtained through the Fourier filter. Secondly, the superiority of wavelet filtering for filtering non-stationary signals is used. By introducing the wavelet transform formula, the filter array expression is established. By judging the size of the two-dimensional fuzzy index of the image fuzziness and the two-dimensional fuzzy threshold of the image fuzziness, the filtering method for different parts of the image is determined according to the filter array expression, thereby improving the accuracy of the image data. Finally, by obtaining the optimized image data input feature matrix, based on the neural network model, the newly input filter array gray value input matrix is ​​judged to identify whether there are hidden dangers and defects in the transmission lines and related equipment, thereby effectively improving the recognition ability of the neural network model and the accuracy of transmission hidden dangers and defects identification, and timely making abnormal alarms and taking emergency measures. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 A flow chart of a high-precision power transmission hidden danger and defect identification method of the present invention;

[0062] Figure 2 The present invention extracts pixel information of an image, performs matrix processing on the image, and converts the spatial domain signal of the image into a frequency domain signal flow chart;

[0063] Figure 3 This is a flow chart of the present invention for identifying whether the image clarity requirement is met when collecting images of power transmission lines and related equipment based on the two-dimensional fuzziness index of the image fuzziness;

[0064] Figure 4 The present invention is to denoise and optimize the image data according to the filter array expression, and determine the image data input feature matrix flow chart. DETAILED DESCRIPTION

[0065] The following description is used to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments described below are only examples, and those skilled in the art may think of other obvious variations.

[0066] Reference Figure 1 As shown, a high-precision power transmission hidden danger and defect identification method includes:

[0067] Based on big data, obtain abnormal situation pictures of related equipment and devices during the installation of transmission lines;

[0068] Extract image pixel information, perform matrix processing on the image, and convert the image's spatial domain signal into a frequency domain signal;

[0069] Based on the two-dimensional fuzziness index of the image fuzziness, it is identified whether the image clarity requirements are met when collecting images of the transmission lines and related equipment. If so, the requirements are met. If not, the collection angle and accuracy of the detection equipment need to be adjusted according to the actual situation;

[0070] According to the filter array expression, the image data is denoised and optimized to determine the input feature matrix of the image data;

[0071] According to the feature matrix input by the image data, based on the neural network model, it is identified whether there are hidden dangers and defects in the transmission lines and related equipment.

[0072] It can be explained that by filtering and analyzing the abnormal conditions of the collected pictures of the relevant equipment and devices during the installation of the transmission line, the noise pollution caused by jitter, movement or external environmental factors can be reduced. Among them, this scheme uses the Fourier transform formula to convert the two-dimensional spatial information of the picture data into a frequency domain signal that is easy to filter. At the same time, the two-dimensional fuzzy index of the picture blur is set according to the transformed frequency domain signal to judge whether the current signal is stable and whether better picture data information can be obtained through the Fourier filter. Secondly, the superiority of wavelet filtering for filtering non-stationary signals is utilized. By introducing the wavelet transform formula, a filter array expression is established. By judging the size of the two-dimensional fuzzy index of the picture blur and the two-dimensional fuzzy threshold of the picture blur, according to the filter array expression, the filtering method for different parts of the picture is determined, thereby improving the accuracy of the picture data. Finally, by obtaining the optimized picture data input feature matrix, based on the neural network model, the gray value input matrix of the newly input filter array is judged to identify whether there are hidden dangers and defects in the transmission line and related equipment.

[0073] Reference Figure 2 As shown, the extracting of picture pixel information, matrix processing of the picture, and converting the spatial domain signal of the picture into a frequency domain signal specifically includes:

[0074] Get the image pixel information and grayscale the red, green, and blue components of each pixel;

[0075] Based on the Gaussian filtering algorithm, the gray value of the Gaussian filtered origin of the n×n window coordinates is calculated;

[0076] According to the gray value after Gaussian filtering, set the n×n gray value input matrix;

[0077] According to the Fourier transform formula, the spatial n×n grayscale value input matrix and the frequency domain n×n grayscale value input matrix are interchanged;

[0078] According to the n×n grayscale value input matrix in the frequency domain, the noise value in the stable state is filtered out based on the components of the image data in the frequency domain;

[0079] The gray value expression after calculating the Gaussian filter of the n×n window coordinate origin is:

[0080]

[0081] Where f(x,y) is the grayscale value of the image that obeys the two-dimensional Gaussian function, g(i,j) is the grayscale value of the image after Gaussian filtering in the i-th row and j-th column, (x0,y0) is the coordinate origin in the n×n window, (x i ,y j ) is the coordinate information of the grayscale value of the image corresponding to the i-th row and j-th column, σ is the standard deviation, p(i,j) is the grayscale value probability of the i-th row and j-th column image, and n is the number of grayscale values;

[0082] The n×n grayscale value input matrix expression is:

[0083]

[0084] Where G(i,j) is the n×n gray value matrix of the image after Gaussian filtering;

[0085] The conversion formula of the spatial input matrix into the frequency domain input matrix is:

[0086]

[0087] Where F(u,v) is the gray value matrix in the frequency domain, n is the number of gray values, and G(i,j) is the n×n gray value matrix of the image after Gaussian filtering.

[0088] It can be explained that when processing image information, it is necessary to grayscale the pixel values ​​at each point and perform preliminary noise reduction using Gaussian filtering. This is both a preliminary cleaning of the image data to ensure that the spatial domain signal of the image data remains intact and standardized, so as to improve the filtering effect of subsequent Fourier filtering and wavelet filtering. Secondly, by performing Fourier transform on the cleaned image data, the spatial domain signal is converted into a frequency domain signal. Through the different components of the frequency domain signal, the noise value in a steady state can be effectively filtered out, thereby achieving a filtering effect in a steady state.

[0089] Reference Figure 3 As shown, the identification of whether the image clarity requirement is met when collecting images of the power transmission line and related equipment based on the two-dimensional fuzziness index of the image fuzziness specifically includes:

[0090] Obtain the standard grayscale value range of the same position images of the same transmission line and related equipment collected at different speeds and angles;

[0091] According to the standard gray value interval, set the two-dimensional blur threshold of the image blur;

[0092] Based on the gray value matrix in the frequency domain, determine the two-dimensional blur index of the image blur;

[0093] According to the comparison result of the two-dimensional blur index of the image blur and the two-dimensional blur threshold of the image blur, it is identified whether the image clarity requirements are met when collecting images of the transmission line and related equipment, and whether the two-dimensional blur index is less than the two-dimensional blur threshold. If so, the requirements are met. If not, the collection angle and accuracy of the detection equipment need to be adjusted according to the actual situation;

[0094] The two-dimensional fuzziness index expression of the image fuzziness is:

[0095]

[0096] Where S is the two-dimensional fuzzy index of the image fuzziness, β is the two-dimensional fuzzy deviation factor of the image fuzziness, is a constant term, N is the number of standard grayscale values ​​of the same position image collected at different speeds and angles for the same transmission line and related equipment, and F(u,v) is the two-dimensional grayscale value in the frequency domain.

[0097] It can be explained that a two-dimensional fuzzy index of image blur is set according to the transformed frequency domain signal to determine whether the current signal is stable and whether better image data information can be obtained through the Fourier filter. Among them, the degree of deviation of images of the same position collected at different speeds and angles on the same transmission line and related equipment is determined by adopting the mean and variance method, thereby establishing a two-dimensional fuzzy index of image blur. By comparing the two-dimensional fuzzy index of image blur with the two-dimensional fuzzy threshold of image blur, it is determined whether the image clarity requirements are met when collecting images of transmission lines and related equipment, and whether the two-dimensional fuzzy index is less than the two-dimensional fuzzy threshold. If so, the requirements are met. If not, the acquisition angle and accuracy of the detection equipment need to be adjusted according to the actual situation, so as to effectively judge and adjust the subsequent filtering accuracy.

[0098] Reference Figure 4 As shown, the denoising and optimization of the image data according to the filter array expression and determining the image data input feature matrix specifically include:

[0099] According to the actual situation of the load device, install the array image acquisition equipment;

[0100] According to the two-dimensional wavelet transform formula, the collected images are denoised and optimized in a non-stationary state;

[0101] According to the judgment result of the two-dimensional blur index of the image blur, the credibility weight is set and the filter array expression is established;

[0102] According to the filter array expression, determine the image data input feature matrix;

[0103] The two-dimensional wavelet transform formula is:

[0104]

[0105] Where WT(α,τ) is the input matrix of wavelet transform for n×n grayscale value input matrix, α and τ are the variable parameters after wavelet transform; G(i,j) is the n×n grayscale value input matrix, (x,y) is the coordinate information of the corresponding image grayscale value;

[0106] The filter array expression is:

[0107]

[0108] Where Z is the grayscale value input matrix of the filter array, S is the two-dimensional fuzzy index matrix of the image fuzziness, m is the number of array image acquisition devices, and F(u,v) and WT(α,τ) are the Fourier filtering results and wavelet filtering results of the image data information collected at the same position, respectively.

[0109] It can be explained that by using array image acquisition equipment to collect multiple groups of image information at the same position, and by averaging multiple groups of data at the same point, the error impact of single image data is reduced and the quality of image data is improved. Secondly, the superiority of wavelet filtering in filtering non-stationary signals is utilized. By introducing the wavelet transform formula, a filter array expression is established, and by judging the size of the two-dimensional blur index of the image blur and the two-dimensional blur threshold of the image blur, the filtering method for different parts of the image is determined according to the filter array expression, thereby improving the accuracy of the image data and the high-quality image data input feature matrix.

[0110] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions only describe the principles of the present invention. The present invention may be subject to various changes and improvements without departing from the spirit and scope of the present invention. These changes and improvements fall within the scope of the present invention. The scope of protection claimed by the present invention is defined by the attached claims and their equivalents.

Claims

1. A high-precision power transmission hidden danger and defect identification method, characterized in that: include: Based on big data, obtain abnormal situation pictures of related equipment and devices during the installation of transmission lines; Extract image pixel information, perform matrix processing on the image, and convert the image's spatial domain signal into a frequency domain signal; Based on the two-dimensional fuzziness index of the image fuzziness, it is identified whether the image clarity requirements are met when collecting images of the transmission lines and related equipment. If so, the requirements are met. If not, the collection angle and accuracy of the detection equipment need to be adjusted according to the actual situation; According to the filter array expression, the image data is denoised and optimized to determine the input feature matrix of the image data; According to the feature matrix input by the image data, based on the neural network model, it is identified whether there are hidden dangers and defects in the transmission lines and related equipment.

2. A high-precision power transmission hidden danger and defect identification method according to claim 1, characterized in that: The step of extracting pixel information of the image, performing matrix processing on the image, and converting the spatial domain signal of the image into a frequency domain signal specifically includes: Get the image pixel information and grayscale the red, green, and blue components of each pixel; Based on the Gaussian filtering algorithm, the gray value of the Gaussian filtered origin of the n×n window coordinates is calculated; According to the gray value after Gaussian filtering, set the n×n gray value input matrix; According to the Fourier transform formula, the spatial n×n grayscale value input matrix and the frequency domain n×n grayscale value input matrix are interchanged; According to the n×n grayscale value input matrix in the frequency domain, based on the components of the image data in the frequency domain, the noise value in the stable state is filtered out; The gray value expression after calculating the Gaussian filter of the n×n window coordinate origin is: Where f(x,y) is the grayscale value of the image that obeys the two-dimensional Gaussian function, g(i,j) is the grayscale value of the image after Gaussian filtering in the i-th row and j-th column, (x0,y0) is the coordinate origin in the n×n window, (x i ,y j ) is the coordinate information of the grayscale value of the image corresponding to the i-th row and j-th column, σ is the standard deviation, p(i,j) is the grayscale value probability of the i-th row and j-th column image, and n is the number of grayscale values; The n×n grayscale value input matrix expression is: Where G(i,j) is the n×n gray value matrix of the image after Gaussian filtering; The conversion formula of the spatial input matrix into the frequency domain input matrix is: Where F(u,v) is the gray value matrix in the frequency domain, n is the number of gray values, and G(i,j) is the n×n gray value matrix of the image after Gaussian filtering.

3. A high-precision power transmission hidden danger and defect identification method according to claim 2, characterized in that: The identification of whether the image clarity requirement is met when collecting images of the power transmission line and related equipment based on the two-dimensional fuzziness index of the image fuzziness specifically includes: Obtain the standard grayscale value range of the same position images of the same transmission line and related equipment collected at different speeds and angles; According to the standard gray value interval, set the two-dimensional blur threshold of the image blur; Based on the gray value matrix in the frequency domain, determine the two-dimensional blur index of the image blur; According to the comparison result of the two-dimensional blur index of the image blur and the two-dimensional blur threshold of the image blur, it is identified whether the image clarity requirements are met when collecting images of the transmission line and related equipment, and whether the two-dimensional blur index is less than the two-dimensional blur threshold. If so, the requirements are met. If not, the collection angle and accuracy of the detection equipment need to be adjusted according to the actual situation; The two-dimensional fuzziness index expression of the image fuzziness is: Where S is the two-dimensional fuzzy index of the image fuzziness, β is the two-dimensional fuzzy deviation factor of the image fuzziness, is a constant term, N is the number of standard grayscale values ​​of the same position image collected at different speeds and angles for the same transmission line and related equipment, and F(u,v) is the two-dimensional grayscale value in the frequency domain.

4. A high-precision power transmission hidden danger and defect identification method according to claim 3, characterized in that: Denoising and optimizing the image data according to the filter array expression to determine the image data input feature matrix specifically includes: According to the actual situation of the load device, install the array image acquisition equipment; According to the two-dimensional wavelet transform formula, the collected images are denoised and optimized in a non-stationary state; According to the judgment result of the two-dimensional blur index of the image blur, the credibility weight is set and the filter array expression is established; According to the filter array expression, determine the image data input feature matrix; The two-dimensional wavelet transform formula is: Where WT(α,τ) is the input matrix of wavelet transform for n×n grayscale value input matrix, α and τ are the variable parameters after wavelet transform; G(i,j) is the n×n grayscale value input matrix, (x,y) is the coordinate information of the corresponding image grayscale value; The filter array expression is: Where Z is the grayscale value input matrix of the filter array, S is the two-dimensional fuzzy index matrix of the image fuzziness, m is the number of array image acquisition devices, and F(u,v) and WT(α,τ) are the Fourier filtering results and wavelet filtering results of the image data information collected at the same position, respectively.

5. A high-precision power transmission hidden danger and defect identification method according to claim 4, characterized in that: The input of the feature matrix according to the image data and the identification of hidden dangers and defects of the power transmission lines and related equipment based on the neural network model specifically include: According to the gray value input matrix formula of the filter array, the characteristic information of the abnormal situation pictures of the relevant equipment and devices in the process of erecting different transmission lines is extracted; According to the characteristic information of the abnormal situation pictures, a characteristic information database of the abnormal situation pictures is established; Based on the neural network model, the grayscale value input matrix of the filter array in the feature information library of the abnormal situation picture is trained and learned; According to the trained neural network model, the newly input filter array grayscale value input matrix is ​​judged to identify whether there are hidden dangers and defects in the transmission line and related equipment. If so, it means that the line and equipment are operating normally. If not, it means that the line and equipment have hidden dangers and defects. By comparing the abnormal type of the feature information library of the abnormal situation picture, a problem warning is sent.

6. A high-precision power transmission hidden danger and defect identification system, used to implement the high-precision power transmission hidden danger and defect identification method as described in any one of claims 1 to 5, characterized in that: include: A data acquisition module, which is used to obtain abnormal situation pictures of related equipment and devices during the installation of power transmission lines based on big data; A data processing module is used to extract pixel information of the image, perform matrix processing on the image, and convert the spatial domain signal of the image into a frequency domain signal; based on the two-dimensional fuzziness index of the image fuzziness, it is used to identify whether the image clarity requirements are met when collecting images of the transmission line and related equipment. If so, the requirements are met; if not, the collection angle and accuracy of the detection equipment need to be adjusted according to the actual situation; According to the filter array expression, the image data is denoised and optimized to determine the input feature matrix of the image data; The anomaly detection module is used to input a feature matrix according to the image data and identify whether there are hidden dangers and defects in the transmission line and related equipment based on the neural network model.

7. A high-precision power transmission hidden danger and defect identification system according to claim 6, characterized in that: The data processing module specifically includes: A signal conversion unit, the signal conversion unit is used to extract pixel information of the picture, perform matrix processing on the picture, and convert the spatial domain signal of the picture into a frequency domain signal; A fuzzy index unit, which is used to identify whether the image clarity requirements are met when collecting images of power transmission lines and related equipment based on a two-dimensional fuzzy index of image fuzziness. If so, the requirements are met; if not, the collection angle and accuracy of the detection equipment need to be adjusted according to actual conditions; An input matrix unit is used to denoise and optimize the image data according to a filter array expression to determine an input feature matrix of the image data.

Images