Power equipment state real-time identification method and system based on artificial intelligence

Abstract

The invention provides a power equipment state real-time identification method and system based on artificial intelligence. The method comprises the steps of collecting original point cloud data of target power equipment in real time based on laser radar scanning equipment; performing point cloud optimization on the original point cloud data to obtain target point cloud data, and performing three-dimensional reconstruction based on the target point cloud data to obtain target three-dimensional model features of the target power equipment; inputting the target three-dimensional model features into a power equipment feature recognition model to obtain a feature recognition result output by the power equipment feature recognition model; and determining the power equipment state of the target power equipment based on the feature recognition result. According to the method, the point cloud data of the power equipment is collected in real time through the laser radar scanning equipment, the recognition real-time performance of the state of the power equipment is improved, the laser radar scanning equipment can adapt to various environments, the state of the power equipment is recognized in combination with the artificial intelligence power equipment feature recognition model, and the recognition accuracy of the state of the power equipment is improved.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a method and system for real-time identification of power equipment status based on artificial intelligence. Background Art

[0002] In the power system, the equipment status of each power equipment will affect the stability of the power system. Therefore, it is crucial to accurately identify the equipment status of the power equipment. Traditional power equipment status identification methods mainly include regular preventive test methods and sensor monitoring methods.

[0003] The regular preventive test method is to conduct offline tests on power equipment regularly. However, the test cycle is fixed, and the equipment status may change suddenly between two tests and cannot be detected in time, resulting in the inability to identify the power equipment status in real time. The sensor monitoring method is to install sensors around the power equipment and obtain equipment parameters through the sensors to identify the equipment status. However, the stability of the sensor is affected by environmental factors, resulting in deviations in the collected equipment parameters, resulting in misidentification or missed identification of the power equipment status, and errors in the identification of the power equipment status. Summary of the invention

[0004] The present invention provides a method and system for real-time identification of power equipment status based on artificial intelligence, so as to improve the real-time identification and identification accuracy of the power equipment status.

[0005] In a first aspect, the present invention provides a method for real-time identification of power equipment status based on artificial intelligence, comprising:

[0006] Based on the laser radar scanning device, the original point cloud data of the target power equipment is collected in real time; the target power equipment is any power equipment in the power system;

[0007] Performing point cloud optimization on the original point cloud data to obtain target point cloud data, and performing three-dimensional reconstruction based on the target point cloud data to obtain target three-dimensional model features of the target power equipment;

[0008] Inputting the target three-dimensional model features into the electric power equipment feature recognition model to obtain the feature recognition results output by the electric power equipment feature recognition model; the electric power equipment feature recognition model is trained based on the sample three-dimensional model features and their corresponding result labels;

[0009] The power device state of the target power device is determined based on the feature recognition result.

[0010] According to the real-time identification method of electric power equipment status based on artificial intelligence provided by an embodiment of the present invention, the determining of the electric power equipment status of the target electric power equipment based on the feature identification result includes:

[0011] If it is determined based on the feature recognition result that the target power device is deformed, acquiring an electrical parameter sensor to collect electrical parameters of the target power device, and determining a power device state of the target power device based on the electrical parameters; or / and

[0012] If it is determined based on the feature recognition result that there is a defect at the connection of the target power device, acquiring the temperature data of the target power device from the temperature sensor, and determining the power device state of the target power device based on the temperature data;

[0013] The electrical parameters include power parameters, current parameters, voltage parameters and partial discharge amount; accordingly, determining the power device state of the target power device based on the electrical parameters includes:

[0014] If the power parameter is greater than the power threshold, and the current parameter is greater than the current threshold, and the partial discharge amount is greater than the discharge amount threshold, and the voltage parameter is less than the voltage threshold, then the state of the power equipment is determined to be a short-circuit fault of the equipment winding;

[0015] If the partial discharge amount is greater than the discharge amount threshold, and the voltage parameter is less than the voltage threshold, determining that the power equipment state is an equipment insulation aging fault;

[0016] If the power parameter is greater than the power threshold, the current parameter is greater than the current threshold, and the partial discharge amount is greater than the discharge amount threshold, then the state of the power equipment is determined to be an equipment winding overheating fault.

[0017] According to the real-time identification method of power equipment status based on artificial intelligence provided by an embodiment of the present invention, the power equipment feature recognition model includes an input layer, multiple intermediate hidden layers and an output layer, the number of nodes in the input layer is the same as the dimension of the target three-dimensional model feature, and the activation functions of the multiple intermediate hidden layers are different;

[0018] Inputting the target three-dimensional model features into the electric power equipment feature recognition model to obtain the feature recognition results output by the electric power equipment feature recognition model includes:

[0019] Inputting the target three-dimensional model features into the input layer of the electric power equipment feature recognition model, processing the target three-dimensional model features based on the first intermediate hidden layer, and outputting a first intermediate result;

[0020] Processing the first intermediate result based on the second intermediate hidden layer, and outputting a second intermediate result;

[0021] Processing the second intermediate result based on the third intermediate hidden layer, and outputting a third intermediate result;

[0022] Processing the third intermediate result based on the output layer to output a first feature recognition result and a second feature recognition result;

[0023] The first feature recognition result represents a predicted probability that the target power device is deformed, and the second feature recognition result represents a predicted probability that a defect exists at a connection of the target power device.

[0024] According to the method for real-time identification of power equipment status based on artificial intelligence provided by an embodiment of the present invention, the processing process of the first intermediate hidden layer is as follows:

[0025] x1=Relu1(W1x0+b1),

[0026] Wherein, x1 represents the first intermediate result, x0 represents the vector corresponding to the target three-dimensional model feature; W1 represents the weight matrix of the first intermediate hidden layer, with a dimension of n*m1, where n is the dimension of the target three-dimensional model feature, and m1 is the number of nodes in the first intermediate hidden layer; b1 represents the bias vector of the first intermediate hidden layer, with a dimension of m1; Relu1(x) represents the activation function of the first intermediate hidden layer;

[0027] The processing of the second intermediate hidden layer is as follows:

[0028]

[0029] Wherein, x2 represents the second intermediate result; W2 represents the weight matrix of the second intermediate hidden layer, the dimension is m1*m2, and m2 is the number of nodes in the second intermediate hidden layer; b2 represents the bias vector of the second intermediate hidden layer, the dimension is m2; Relu2(x) represents the activation function of the second intermediate hidden layer;

[0030] The processing process of the third intermediate hidden layer is as follows:

[0031] x3=Relu3(W3x2+b3),

[0032] Wherein, x3 represents the third intermediate result; W3 represents the weight matrix of the third intermediate hidden layer, with a dimension of m2*m3, where m3 is the number of nodes in the third intermediate hidden layer; b3 represents the bias vector of the third intermediate hidden layer, with a dimension of m3; Relu3(x) represents the activation function of the second intermediate hidden layer;

[0033] The processing of the output layer is as follows:

[0034] y = Relu4(W4x3+b4),

[0035] y represents the feature recognition result, with a dimension of 2; W4 represents the weight matrix of the output layer, with a dimension of m3*2; b4 represents the bias vector of the output layer, with a dimension of 2; Relu4(x) represents the activation function of the output layer.

[0036] According to the artificial intelligence-based real-time identification method for power equipment status provided by an embodiment of the present invention, the point cloud optimization of the original point cloud data to obtain target point cloud data includes:

[0037] Filtering the interference point cloud in the original point cloud data to obtain filtered point cloud data;

[0038] The point cloud in the filtered point cloud data is processed to enhance the information feature expression capability to obtain the target point cloud data.

[0039] According to the artificial intelligence-based real-time identification method for power equipment status provided by an embodiment of the present invention, filtering the interference point cloud in the original point cloud data to obtain filtered point cloud data includes:

[0040] For each first target point cloud in the original point cloud data, determine a neighborhood of each first target point cloud and a convex hull volume of the neighborhood of each first target point cloud;

[0041] determining an angular discreteness of each first target point cloud based on a vector between each first target point cloud and each second target point cloud in its immediate neighborhood;

[0042] determining a spherical volume of each first target point cloud based on a distance between each first target point cloud and each second target point cloud in its immediate neighborhood;

[0043] The original point cloud data is filtered based on the convex hull volume of the neighborhood of each first target point cloud, and the angular discreteness and sphere volume of each first target point cloud to obtain the filtered point cloud data.

[0044] According to the method for real-time identification of power equipment status based on artificial intelligence provided by an embodiment of the present invention, the point cloud in the filtered point cloud data is subjected to information feature expression capability enhancement processing to obtain the target point cloud data, including:

[0045] According to the spatial coordinates of each point cloud in the filtered point cloud data, the point clouds in the filtered point cloud data are grouped to obtain a plurality of point cloud subsets;

[0046] Calculate the feature vector between each point cloud subset;

[0047] Determine the correlation features between the point cloud subsets based on the feature vectors of each point cloud subset;

[0048] New attribute information is assigned to the point cloud in each point cloud subset based on the association features between the point cloud subsets to obtain the target point cloud data.

[0049] In a second aspect, the present invention further provides a real-time identification system for power equipment status based on artificial intelligence, which is applied to the real-time identification method for power equipment status based on artificial intelligence as described in the first aspect, and the real-time identification system for power equipment status based on artificial intelligence includes:

[0050] A point cloud acquisition module, used for acquiring original point cloud data of a target power device in real time based on a laser radar scanning device; the target power device is any power device in the power system;

[0051] A point cloud processing module, used to perform point cloud optimization on the original point cloud data to obtain target point cloud data, and perform three-dimensional reconstruction based on the target point cloud data to obtain target three-dimensional model features of the target power equipment;

[0052] A feature recognition module, used for inputting the target three-dimensional model features into the power equipment feature recognition model to obtain the feature recognition results output by the power equipment feature recognition model; the power equipment feature recognition model is trained based on the sample three-dimensional model features and their corresponding result labels;

[0053] The power equipment state identification module is used to determine the power equipment state of the target power equipment based on the feature identification result.

[0054] In a third aspect, the present invention further provides an electronic device, comprising: a memory for storing a computer software program; a processor for reading and executing the computer software program, thereby implementing any of the above-mentioned methods for real-time identification of the state of power equipment based on artificial intelligence.

[0055] In a fourth aspect, the present invention further provides a non-transitory computer-readable storage medium, wherein a computer software program is stored in the storage medium, and when the computer software program is executed by a processor, the method for real-time identification of the state of an electric power device based on artificial intelligence as described above is implemented.

[0056] In a fifth aspect, the present invention further provides a computer program product, comprising a computer program, which, when executed by a processor, implements any of the above-mentioned methods for real-time identification of the state of power equipment based on artificial intelligence.

[0057] The artificial intelligence-based real-time identification method for the status of power equipment provided in the embodiment of the present invention collects point cloud data of the power equipment in real time through a laser radar scanning device, thereby improving the real-time identification of the status of the power equipment, and further reconstructs three-dimensional model features based on the point cloud data to identify the status of the power equipment. Since the laser radar scanning device can be applied to various environments and has strong anti-interference ability to various environments, it can accurately obtain three-dimensional model features in various environments, thereby improving the recognition accuracy of the status of the power equipment. The power equipment feature recognition model pre-trained by artificial intelligence is further combined to perform power equipment status identification, thereby being able to accurately obtain the status of the power equipment, thereby further improving the recognition accuracy of the status of the power equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 It is a flow chart of a method for real-time identification of power equipment status based on artificial intelligence provided by an embodiment of the present invention;

[0059] Figure 2 It is a structural diagram of a real-time identification system for power equipment status based on artificial intelligence provided by an embodiment of the present invention;

[0060] Figure 3 An embodiment diagram of an electronic device provided by an embodiment of the present invention;

[0061] Figure 4 An embodiment diagram of a computer-readable storage medium provided for an embodiment of the present invention. DETAILED DESCRIPTION

[0062] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0063] In the description of the present invention, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0064] In the description of the present invention, the term "for example" is used to mean "used as an example, illustration or explanation". Any embodiment described as "for example" in the present invention is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is given to enable any technician in the field to implement and use the present invention. In the following description, details are listed for the purpose of explanation. It should be understood that a person of ordinary skill in the art can recognize that the present invention can be implemented without using these specific details. In other examples, well-known structures and processes will not be elaborated in detail to avoid obscuring the description of the present invention with unnecessary details. Therefore, the present invention is not intended to be limited to the embodiments shown, but is consistent with the widest scope consistent with the principles and features disclosed in the present invention.

[0065] Optional, see Figure 1 As shown, Figure 1 : is a flow chart of a method for real-time identification of power equipment status based on artificial intelligence provided by the present invention. The execution subject of the method for real-time identification of power equipment status based on artificial intelligence in the embodiment of the present invention is a status identification system. Therefore, the method for real-time identification of power equipment status based on artificial intelligence includes:

[0066] Step 10: Collect the original point cloud data of the target power equipment in real time based on the laser radar scanning device.

[0067] Step 20, performing point cloud optimization on the original point cloud data to obtain target point cloud data, and performing three-dimensional reconstruction based on the target point cloud data to obtain target three-dimensional model features of the target power equipment.

[0068] The state recognition system of the embodiment of the present invention is connected to a laser radar scanning device, which can collect point cloud data of objects, such as power equipment such as transformers, switch cabinets, insulators, disconnectors, circuit breakers, etc. in the power system. Therefore, when identifying the state of the power equipment, the state recognition system calls the laser radar scanning device to scan any target power equipment in the power system in real time, and collects the original point cloud data of the target power equipment.

[0069] Among them, the collected original point cloud data may contain interfering point cloud data, or some point cloud data in the original point cloud data may have weak information feature expression capabilities. Therefore, the state recognition system performs point cloud filtering on the interfering point cloud data in the original point cloud data to obtain filtered point cloud data. Furthermore, the state recognition system performs point cloud optimization on the filtered point cloud data to enhance the information feature expression capabilities of each point cloud to obtain target point cloud data, as described in steps 201 to 202.

[0070] Furthermore, the state recognition system performs three-dimensional reconstruction based on the target point cloud data to obtain the target three-dimensional model features of the target power equipment. Commonly used three-dimensional reconstruction algorithms for three-dimensional reconstruction include those based on geometric algorithms and deep learning algorithms. The specific process of three-dimensional reconstruction will not be described in detail in the embodiments of the present invention.

[0071] Step 30: input the target three-dimensional model features into the electric power equipment feature recognition model to obtain the feature recognition results output by the electric power equipment feature recognition model.

[0072] The state recognition system of the embodiment of the present invention is embedded with a power equipment feature recognition model, which is obtained by training a preset model based on the sample three-dimensional model features of the sample power equipment and its corresponding result labels, wherein the preset model is such as a feedback neural network, a graph network model, a feedforward neural network, etc. For example, the feedback neural network is trained by the sample three-dimensional model features of the sample power equipment and its corresponding result labels in the embodiment of the present invention to obtain the power equipment feature recognition model.

[0073] Therefore, the state recognition system inputs the target three-dimensional model features into the power equipment feature recognition model, processes the target three-dimensional model features through the power equipment feature recognition model, and outputs the feature recognition results of the target three-dimensional model features, as described in steps 301 to 304. Therefore, the state recognition system can obtain the feature recognition results of the target three-dimensional model features.

[0074] Step 40: determine the power device state of the target power device based on the feature recognition result.

[0075] The feature recognition result in the embodiment of the present invention can be understood as the predicted probability of abnormal state of the power equipment. The abnormal state of the power equipment includes deformation of the power equipment and defects in the connection of the power equipment.

[0076] Therefore, the state recognition system determines whether the feature recognition result is greater than or equal to a preset threshold, wherein the preset threshold is set according to actual conditions. If it is determined that the feature recognition result is greater than or equal to the preset threshold, the state recognition system determines that there is an abnormality in the power equipment state of the target power equipment, that is, the power equipment is deformed or / and there are defects in the connection of the power equipment. If it is determined that the feature recognition result is less than the preset threshold, the state recognition system determines that there is no abnormality in the power equipment state of the target power equipment, as described in steps 401 to 402.

[0077] The embodiment of the present invention collects point cloud data of power equipment in real time through a laser radar scanning device, thereby improving the real-time recognition of the power equipment status, and further reconstructs three-dimensional model features based on the point cloud data to identify the power equipment status. Since the laser radar scanning device can be applied to various environments and has strong anti-interference ability to various environments, it can accurately obtain three-dimensional model features in various environments, thereby improving the recognition accuracy of the power equipment status. The power equipment status is further recognized by combining the power equipment feature recognition model pre-trained by artificial intelligence, thereby being able to accurately obtain the power equipment status, further improving the recognition accuracy of the power equipment status.

[0078] In one embodiment, the description of step 201 to step 202 is as follows:

[0079] Step 201 : filtering the interference point cloud in the original point cloud data to obtain filtered point cloud data.

[0080] Specifically, the state recognition system filters the interference point cloud in the original point cloud data to obtain filtered point cloud data, as described in steps 2011 to 2014:

[0081] Step 2011: for each first target point cloud in the original point cloud data, determine the neighborhood of each first target point cloud and the convex hull volume of the neighborhood of each first target point cloud.

[0082] Optionally, for each first target point cloud P in the original point cloud data i , the state recognition system determines each first target point cloud P i The k nearest neighbor domain is obtained by i = {P i1 ,P i2 ,....,P ik}, where k is set according to actual conditions, such as 8, 10, 20, etc. At the same time, each first target point cloud P is determined i The nearest neighbor N i The convex hull volume V i , where the convex hull volume can be calculated using the Quick Hull algorithm.

[0083] Step 2012: determining the angular discreteness of each first target point cloud based on the vectors between each first target point cloud and each second target point cloud in its neighborhood.

[0084] Furthermore, the state recognition system calculates each first target point cloud P i and its neighboring domain N i Each second target point cloud P in ij vector between (j=1,2,...,k)

[0085] Furthermore, the state recognition system calculates each first target point cloud P i and its neighboring domain N i In each second target point cloud, any two vectors and The angle θ between inm , angle θ inm The calculation formula is as follows:

[0086]

[0087] Furthermore, the state recognition system is based on the angle θ between any two vectors. inm Calculate the first target point cloud P i The mean angle of And according to the angle mean Calculate each first target point cloud P i Angular dispersion AD i , degree dispersion AD i The calculation formula is as follows:

[0088]

[0089] in, It represents the number of combinations of selecting 2 elements from k elements.

[0090] Step 2013: determining the spherical volume of each first target point cloud based on the distance between each first target point cloud and each second target point cloud in its neighborhood.

[0091] Furthermore, the state recognition system calculates each first target point cloud P i and its neighboring domain N i Each second target point cloud P in ij The distance between i , can be calculated by the distance formula between two points, which will not be described in detail in the embodiment of the present invention.

[0092] Furthermore, the state recognition system is based on each first target point cloud P i and its neighboring domain N i Each second target point cloud P in ij The distance between i , calculate each first target point cloud P i The mean distance And each first target point cloud P i Centered on the mean distance For radius r, determine each first target point cloud P i The volume of the sphere V all, where the sphere volume calculation formula is a conventional calculation formula and will not be repeated here.

[0093] Step 2014 , filtering the original point cloud data based on the convex hull volume of the neighborhood of each first target point cloud, and the angle discreteness and sphere volume of each first target point cloud to obtain filtered point cloud data.

[0094] Furthermore, the state recognition system is based on each first target point cloud P i The nearest neighbor N i The convex hull volume V i , and each first target point cloud P i The volume of the sphere V all , calculate each first target point cloud P i Neighborhood Volume Ratio NVR i , and the neighborhood volume ratio NVR i Determine for each first target point cloud P i The sparsity of the neighborhood volume ratio NVR i The calculation formula is as follows:

[0095] NVR i =V i / V all ;

[0096] Furthermore, the state recognition system traverses each first target point cloud P in the original point cloud data. i , will satisfy the angle discreteness AD i Less than or equal to the discrete threshold, and the neighborhood volume ratio NVR i The first target point cloud P that is greater than or equal to the volume ratio threshold i The filtered point cloud data set is retained and the angle discrete degree AD i Greater than discrete threshold, or / and, neighborhood volume ratio NVR i The first target point cloud P that is smaller than the volume ratio threshold i Eliminate and obtain filtered point cloud data, where the discrete threshold and volume ratio threshold are set according to actual conditions.

[0097] Step 202 , performing information feature expression capability enhancement processing on the point cloud in the filtered point cloud data to obtain target point cloud data.

[0098] Specifically, the state recognition system performs information feature expression capability enhancement processing on the point cloud in the filtered point cloud data to obtain target point cloud data, as described in steps 2021 to 2024:

[0099] Step 2021, grouping the point clouds in the filtered point cloud data according to the spatial coordinates of each point cloud in the filtered point cloud data to obtain multiple point cloud subsets.

[0100] Optionally, the state recognition system groups the point clouds in the filtered point cloud data according to the spatial distribution characteristics and the spatial coordinates of each point cloud in the filtered point cloud data to obtain multiple point cloud subsets, wherein the spatial distribution characteristics are such as a spatial grid, so it can be understood that the filtered point cloud data in the entire point cloud space is divided into a number of small cubic grid units by a spatial grid division method, and the point cloud in each cubic grid unit forms a subset. In one embodiment, the filtered point cloud data is represented by P = {p i =(x i ,y i ,z i )|i=1,2,...,n}, the side length of the spatial grid is s, where n represents the number of point clouds in the filtered point cloud data, (x i ,y i ,z i ) represents the i-th point cloud p i For the i-th point cloud p in the filtered point cloud data i , the index of the grid cell where it is located (g x ,g y ,g z ) is calculated as follows:

[0101]

[0102] in, Indicates the rounding down operation, (g x ,g y ,g z ) represents the i-th point cloud p i The grid cell numbers in the x-direction, y-direction and z-direction are respectively located. Therefore, the point cloud in the filtered point cloud data can be divided into different grid cells to form several point cloud subsets G = {G j |j=1,2,...,m}, where m represents the number of point cloud subsets, G j represents the point cloud subset within the jth grid cell.

[0103] Step 2022, calculate the feature vector between each point cloud subset.

[0104] Furthermore, for each point cloud subset, the state recognition system calculates a feature vector between each point cloud subset, wherein the feature vector in the embodiment of the present invention includes information such as the distribution center of the point cloud, the degree of dispersion, and the relative relationship between point clouds.

[0105] For the point cloud subset G j , its center of gravity C j =(x cj ,ycj ,z cj ) is calculated as follows:

[0106]

[0107] Among them, |G j | represents the point cloud subset G j The number of inliers.

[0108] The embodiment of the present invention uses a covariance matrix to measure the dispersion of point clouds in a point cloud subset. In one embodiment, the covariance matrix is ​​∑j, and the elements of the covariance matrix are Therefore, the element It can be expressed as:

[0109]

[0110] Among them, p im Represents the point cloud p i The coordinate value on the m axis, p in Represents the point cloud p i The coordinate value on the n-axis, such as when m = x, p ix =x i .

[0111] Therefore, the covariance matrix ∑j can be obtained as:

[0112]

[0113] Furthermore, the state recognition system obtains three solutions by solving the ternary characteristic equation |λI-∑j|=0, and the three solutions are the eigenvalues ​​λ of the covariance matrix ∑j. j1 ,λ j2 ,λ j3 (λ j1 >λ j2 >λ j3 ), where I represents the 3*3 identity matrix.

[0114] Furthermore, the state recognition system uses the ratio of the maximum eigenvalue to the minimum eigenvalue to represent the degree of dispersion, so the degree of dispersion can be expressed as: R j =λ j1 / λ j3 , where the larger the ratio is, the more dispersed the point cloud is distributed in the point cloud subset.

[0115] Furthermore, for p i ∈G j , the state recognition system traverses the point cloud p i In the point cloud subset G j The point cloud p′ of the nearest neighbor ini , point cloud p i With point cloud p′ i The distance between them is denoted as d ii′ . Further, the state recognition system calculates the point cloud subset G j The average nearest neighbor distance d of the midpoint cloud j , the specific calculation formula is as follows:

[0116]

[0117] Therefore, the point cloud subset G j The eigenvector of can be expressed as F j =(xc j ,yc j ,z cj ,R j ,d j ).

[0118] Step 2023: determining the association features between the point cloud subsets based on the feature vector of each point cloud subset.

[0119] Furthermore, the state recognition system determines the correlation features between point cloud subsets based on the feature vector of each point cloud subset, and the correlation features include relative position features and density change features between adjacent subsets.

[0120] Therefore, for two adjacent point cloud subsets G j and G k , calculate the point cloud subset G j And the point cloud subset G k The vector between the centroid coordinates of j And the point cloud subset G k The relative position features between them, therefore, the point cloud subset G j And the point cloud subset G k The relative position characteristics between can be expressed as:

[0121]

[0122] in, Represents the relative position feature, representing the point cloud subset G k The center of gravity points to the point cloud subset G j The direction and distance of the center of gravity; C j Represents the point cloud subset G j The centroid coordinates, C k Represents the point cloud subset G k The centroid coordinates of Relative position feature The component on the x-axis, Relative position feature The component on the y-axis, Relative position feature The component along the z-axis.

[0123] Furthermore, the state recognition system determines the point cloud subset G j The subset density ρ j And the point cloud subset G k The subset density ρ k , and according to the subset density ρ j and subset density ρ k Calculate the density change rate Δρ jk , the density change rate Δρ jk That is the point cloud subset G j And the point cloud subset G k The density change characteristics between jk The calculation formula is as follows:

[0124]

[0125] Among them, V j Represents the volume of a grid cell.

[0126] Therefore, the correlation characteristics between point cloud subsets are

[0127] Step 2024: assign new attribute information to the point cloud in each point cloud subset based on the association features between the point cloud subsets to obtain target point cloud data.

[0128] Furthermore, the correlation features between the point cloud subsets of the state recognition system are used to assign new attribute information to the point cloud in each point cloud subset to obtain the target point cloud data. i ∈G j , the feature vector F in the subset of the point cloud subset where it is located j and the associated features with adjacent subsets (with point cloud p i The associated features corresponding to the grid cells adjacent to the grid cell are fused to obtain a new point cloud representation Among them, A j1 ,A j2 ,... represents the point cloud subset G j Adjacent associated features, all point clouds are processed as above to obtain the target point cloud data.

[0129] The embodiment of the present invention filters the interference point cloud in the original point cloud data, thereby reducing the influence of environmental factors. At the same time, the point cloud in the filtered point cloud data is enhanced in information feature expression capability, so that the three-dimensional model features can be accurately reconstructed in the end. Therefore, the three-dimensional model features in various environments can be accurately obtained, thereby improving the recognition accuracy of the power equipment status.

[0130] In one embodiment, the electric power equipment feature recognition model includes an input layer, multiple intermediate hidden layers and an output layer, the number of nodes in the input layer is the same as the dimension of the target three-dimensional model feature, and the activation functions of the multiple intermediate hidden layers are different. The intermediate hidden layers in the embodiment of the present invention include a first intermediate hidden layer, a second intermediate hidden layer and a third intermediate hidden layer, wherein the number of nodes in the first intermediate hidden layer is m1, the number of nodes in the second intermediate hidden layer is m2, the number of nodes in the third intermediate hidden layer is m3, and the dimension of the target three-dimensional model feature is n, therefore, the description of steps 301 to 304 is as follows:

[0131] Step 301, input the target three-dimensional model features into the input layer of the power equipment feature recognition model, process the target three-dimensional model features based on the first intermediate hidden layer, and output a first intermediate result; Step 302, process the first intermediate result based on the second intermediate hidden layer, and output a second intermediate result; Step 303, process the second intermediate result based on the third intermediate hidden layer, and output a third intermediate result.

[0132] Specifically, the state recognition system inputs the target three-dimensional model features into the input layer of the power equipment feature recognition model. The input layer receives the input target three-dimensional model features, does not perform any processing on the target three-dimensional model features, and directly outputs the target three-dimensional model features to the first intermediate hidden layer.

[0133] Furthermore, the first intermediate hidden layer processes the target three-dimensional model features and outputs a first intermediate result to the second intermediate hidden layer, wherein the processing process of the first intermediate hidden layer is as follows:

[0134]

[0135] Among them, x1 represents the first intermediate result, x0 represents the vector corresponding to the target three-dimensional model feature; W1 represents the weight matrix of the first intermediate hidden layer, with the dimension n*m1, n is the dimension of the target three-dimensional model feature, and m1 is the number of nodes in the first intermediate hidden layer; b1 represents the bias vector of the first intermediate hidden layer, with the dimension m1; Relu1(x) represents the activation function of the first intermediate hidden layer.

[0136] Furthermore, the second intermediate hidden layer processes the first intermediate result and outputs the second intermediate result to the third intermediate hidden layer. The processing process of the second intermediate hidden layer is as follows:

[0137]

[0138] Among them, x2 represents the second intermediate result; W2 represents the weight matrix of the second intermediate hidden layer, with dimension m1*m2, where m2 is the number of nodes in the second intermediate hidden layer; b2 represents the bias vector of the second intermediate hidden layer, with dimension m2; Relu2(x) represents the activation function of the second intermediate hidden layer.

[0139] Furthermore, the third intermediate hidden layer processes the second intermediate result and outputs the third intermediate result to the output layer. The processing process of the third intermediate hidden layer is as follows:

[0140]

[0141] Among them, x3 represents the third intermediate result; W3 represents the weight matrix of the third intermediate hidden layer, with dimension m2*m3, and m3 is the number of nodes in the third intermediate hidden layer; b3 represents the bias vector of the third intermediate hidden layer, with dimension m3; Relu3(x) represents the activation function of the second intermediate hidden layer.

[0142] Step 304: Process the third intermediate result based on the output layer, and output the first feature recognition result and the second feature recognition result.

[0143] Furthermore, in the embodiment of the present invention, the output layer has two nodes, which respectively correspond to the feature recognition result of deformation of the target power equipment and the feature recognition result of defects at the connection of the target power equipment.

[0144] Therefore, the output layer processes the third intermediate result and outputs the first feature recognition result and the second feature recognition result, wherein the first feature recognition result represents the predicted probability of deformation of the target power equipment, and the second feature recognition result represents the predicted probability of defects at the connection of the target power equipment. The processing process of the output layer is as follows:

[0145]

[0146] y represents the feature recognition result, with a dimension of 2; W4 represents the weight matrix of the output layer, with a dimension of m3*2; b4 represents the bias vector of the output layer, with a dimension of 2; Relu4(x) represents the activation function of the output layer.

[0147] Furthermore, for the training phase of the power equipment feature recognition model, the sample three-dimensional model features and their corresponding result labels are used as input data and target data. Each time the three-dimensional model features of a sample are input, the predicted feature recognition result (i.e., the output y of the output layer) is obtained through the forward propagation process of the above model, and then the weight and bias of the model are adjusted according to the difference between the predicted result and the true result label, so that the model can gradually learn the correct feature recognition mode, wherein the adjustment process is implemented through the back propagation algorithm, and the parameters are updated according to the gradient of the loss function to the model parameters.

[0148] In one embodiment, for a sample, the true result label is t (the dimension is 2, t1 represents the true label of the target power equipment being deformed, and t2 represents the true label of the target power equipment having a defect at the connection, and the value is 0 or 1), and the model prediction result is y (the dimension is 2, such as the predicted probability of the two features calculated by the output layer above). Therefore, the loss function of the power equipment feature recognition model is expressed as:

[0149]

[0150] Among them, L(y,t) represents the loss function, (y i -t i ) 2 represents the mean square error term, which measures the square difference between the predicted value and the true value, making the predicted result as close to the true result as possible; log(1+|y i -t i |) represents the logarithmic term, which gives a greater penalty to larger errors, prompting the model to pay more attention to samples with inaccurate predictions, better optimize the model parameters, and improve the generalization ability of the model.

[0151] The embodiment of the present invention recognizes the state of the electric device through the electric device feature recognition model, so the state of the electric device can be accurately obtained, thereby improving the recognition accuracy of the state of the electric device.

[0152] In one embodiment, in order to more accurately identify the state of the power equipment, it is necessary to combine the parameters collected by the sensor and the feature recognition results for joint recognition. Therefore, the description of step 401 to step 402 is as follows:

[0153] Step 401: if it is determined based on the feature recognition result that the target power device is deformed, an electrical parameter sensor is used to collect electrical parameters of the target power device, and an electrical device state of the target power device is determined based on the electrical parameters; or / and

[0154] Step 402: If it is determined based on the feature recognition result that there is a defect at the connection of the target power device, a temperature sensor is used to collect temperature data of the target power device, and a power device state of the target power device is determined based on the temperature data.

[0155] The feature recognition result in the embodiment of the present invention includes a first feature recognition result and a second feature recognition result, wherein the first feature recognition result represents the predicted probability of deformation of the target power equipment, and the second feature recognition result represents the predicted probability of defects at the connection of the target power equipment. Therefore, if the first feature recognition result is greater than or equal to the first preset threshold, it is determined that the target power equipment is deformed. If the second feature recognition result is greater than or equal to the second preset threshold, it is determined that there is a defect at the connection of the target power equipment, wherein the first preset threshold and the second preset threshold are set according to actual conditions.

[0156] Specifically, if it is determined based on the feature recognition results that the target power equipment has been deformed, the state recognition system obtains the electrical parameters of the target power equipment collected by the electrical parameter sensor, where the electrical parameter sensors include current sensors, voltage sensors, power sensors and discharge sensors. Therefore, the electrical parameters include power parameters, current parameters, voltage parameters and local discharge amounts.

[0157] Further, the electrical parameters of the state identification system determine the state of the target power equipment, specifically: if the power parameter is greater than the power threshold, and the current parameter is greater than the current threshold, and the partial discharge is greater than the discharge threshold, and the voltage parameter is less than the voltage threshold, then the state of the power equipment is determined to be a short-circuit fault in the equipment winding. If the partial discharge is greater than the discharge threshold, and the voltage parameter is less than the voltage threshold, then the state of the power equipment is determined to be an insulation aging fault in the equipment; if the power parameter is greater than the power threshold, and the current parameter is greater than the current threshold, and the partial discharge is greater than the discharge threshold, then the state of the power equipment is determined to be an overheating fault in the equipment winding, wherein, for different target power equipment, the power threshold, current threshold, discharge threshold and voltage threshold are different, and are set according to actual conditions.

[0158] In one embodiment, the target power equipment is a transformer. If the transformer box is determined to have local expansion and deformation according to the first feature recognition result, if the current sensor detects that the current exceeds the threshold of the normal operating value, and the discharge sensor detects an increase in the local discharge, it can be determined that the transformer winding has a short circuit fault. If the surface of the transformer bushing is determined to be corroded, that is, deformed, according to the first feature recognition result, if the voltage sensor monitors abnormal voltage fluctuations at both ends of the transformer bushing, and the discharge sensor detects that the local discharge exceeds the normal range, it can be determined that the transformer bushing has an insulation fault.

[0159] In one embodiment, the target power equipment is a switch cabinet. If it is determined according to the first feature recognition result that the contact part of the switch cabinet has expanded and deformed, if the current sensor detects that the current passing through the switch cabinet contact is unstable at this time, and the partial discharge amount increases near the switch cabinet contact, it can be determined that the switch cabinet contact has an overheating fault. If it is determined according to the first feature recognition result that cracks appear on the surface of the switch cabinet insulation layer, that is, deformation occurs, if the voltage sensor monitors that the bus voltage is abnormal at this time, and the discharge amount sensor detects that the partial discharge signal is enhanced, it can be determined that the bus of the switch cabinet has an insulation damage fault.

[0160] Furthermore, if it is determined based on the feature recognition result that there is a defect at the connection of the target power equipment, the state recognition system obtains the temperature data of the target power equipment collected by the temperature sensor, and determines the power equipment state of the target power equipment based on the temperature data, specifically: if the temperature data is greater than or equal to the preset temperature threshold, the power equipment state of the target power equipment is determined to be a poor contact fault, wherein the preset temperature threshold is set according to the actual situation, such as 30 degrees, 35 degrees, etc.

[0161] The embodiment of the present invention combines the parameters collected by the sensor with the feature recognition results to jointly identify the state of the power equipment, which can more accurately identify the state of the power equipment and improve the recognition accuracy of the state of the power equipment.

[0162] Furthermore, the artificial intelligence-based real-time identification system for the status of electric equipment provided by the present invention is described below. The artificial intelligence-based real-time identification system for the status of electric equipment described below and the artificial intelligence-based real-time identification method for the status of electric equipment described above can be referred to each other.

[0163] Optional, see Figure 2 , Figure 2 The present invention provides a schematic diagram of the structure of a real-time identification system for power equipment status based on artificial intelligence, which includes:

[0164] Point cloud acquisition module 210, used for real-time acquisition of original point cloud data of target power equipment based on laser radar scanning equipment; the target power equipment is any power equipment in the power system;

[0165] The point cloud processing module 220 is used to optimize the original point cloud data to obtain target point cloud data, and perform three-dimensional reconstruction based on the target point cloud data to obtain target three-dimensional model features of the target power equipment;

[0166] The feature recognition module 230 is used to input the target three-dimensional model features into the power equipment feature recognition model to obtain the feature recognition results output by the power equipment feature recognition model; the power equipment feature recognition model is trained based on the sample three-dimensional model features and their corresponding result labels;

[0167] The power device state identification module 240 is used to determine the power device state of the target power device based on the feature identification result.

[0168] The embodiment of the present invention collects point cloud data of power equipment in real time through a laser radar scanning device, thereby improving the real-time recognition of the power equipment status, and further reconstructs three-dimensional model features based on the point cloud data to identify the power equipment status. Since the laser radar scanning device can be applied to various environments and has strong anti-interference ability to various environments, it can accurately obtain three-dimensional model features in various environments, thereby improving the recognition accuracy of the power equipment status. The power equipment status is further recognized by combining the power equipment feature recognition model pre-trained by artificial intelligence, thereby being able to accurately obtain the power equipment status, further improving the recognition accuracy of the power equipment status.

[0169] See also Figure 3 , Figure 3 FIG. 1 is an embodiment diagram of an electronic device provided by an embodiment of the present invention. Figure 3 As shown, an embodiment of the present invention provides an electronic device 300, including a memory 310, a processor 320, and a computer program 311 stored in the memory 310 and executable on the processor 320. When the processor 320 executes the computer program 311, the following steps are implemented:

[0170] The original point cloud data of the target power equipment is collected in real time based on the laser radar scanning device; the target power equipment is any power equipment in the power system;

[0171] Perform point cloud optimization on the original point cloud data to obtain target point cloud data, and perform three-dimensional reconstruction based on the target point cloud data to obtain target three-dimensional model features of the target power equipment;

[0172] Input the target three-dimensional model features into the power equipment feature recognition model to obtain the feature recognition results output by the power equipment feature recognition model; the power equipment feature recognition model is trained based on the sample three-dimensional model features and their corresponding result labels;

[0173] The power device state of the target power device is determined based on the feature recognition result.

[0174] See also Figure 4 , Figure 4 Detailed description of an embodiment of a computer-readable storage medium provided by an embodiment of the present invention. Figure 4 As shown, this embodiment provides a computer-readable storage medium 400, on which a computer program 311 is stored. When the computer program 311 is executed by a processor, the following steps are implemented:

[0175] The original point cloud data of the target power equipment is collected in real time based on the laser radar scanning device; the target power equipment is any power equipment in the power system;

[0176] Perform point cloud optimization on the original point cloud data to obtain target point cloud data, and perform three-dimensional reconstruction based on the target point cloud data to obtain target three-dimensional model features of the target power equipment;

[0177] Input the target three-dimensional model features into the power equipment feature recognition model to obtain the feature recognition results output by the power equipment feature recognition model; the power equipment feature recognition model is trained based on the sample three-dimensional model features and their corresponding result labels;

[0178] The power device state of the target power device is determined based on the feature recognition result.

[0179] On the other hand, the present invention further provides a computer program product, the computer program product includes a computer program, the computer program can be stored in a non-transitory computer-readable storage medium, when the computer program is executed by a processor, the computer can execute the real-time identification method of the power equipment state based on artificial intelligence provided by the above methods, the method includes:

[0180] The original point cloud data of the target power equipment is collected in real time based on the laser radar scanning device; the target power equipment is any power equipment in the power system;

[0181] Perform point cloud optimization on the original point cloud data to obtain target point cloud data, and perform three-dimensional reconstruction based on the target point cloud data to obtain target three-dimensional model features of the target power equipment;

[0182] Input the target three-dimensional model features into the power equipment feature recognition model to obtain the feature recognition results output by the power equipment feature recognition model; the power equipment feature recognition model is trained based on the sample three-dimensional model features and their corresponding result labels;

[0183] The power device state of the target power device is determined based on the feature recognition result.

[0184] The system embodiment described above is merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art may understand and implement it without creative work.

[0185] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0186] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for real-time identification of power equipment status based on artificial intelligence, characterized in that: include: Real-time collection of original point cloud data of target power equipment based on LiDAR scanning equipment; The target power equipment is any power equipment in the power system; Performing point cloud optimization on the original point cloud data to obtain target point cloud data, and performing three-dimensional reconstruction based on the target point cloud data to obtain target three-dimensional model features of the target power equipment; Inputting the target three-dimensional model features into the electric power equipment feature recognition model to obtain the feature recognition results output by the electric power equipment feature recognition model; the electric power equipment feature recognition model is trained based on the sample three-dimensional model features and their corresponding result labels; The power device state of the target power device is determined based on the feature recognition result.

2. The method for real-time identification of power equipment status based on artificial intelligence according to claim 1 is characterized in that: The determining the power device state of the target power device based on the feature recognition result includes: If it is determined based on the feature recognition result that the target power device is deformed, acquiring an electrical parameter sensor to collect electrical parameters of the target power device, and determining a power device state of the target power device based on the electrical parameters; or / and If it is determined based on the feature recognition result that there is a defect at the connection of the target power device, acquiring the temperature data of the target power device from the temperature sensor, and determining the power device state of the target power device based on the temperature data; The electrical parameters include power parameters, current parameters, voltage parameters and partial discharge amount; accordingly, determining the power device state of the target power device based on the electrical parameters includes: If the power parameter is greater than the power threshold, and the current parameter is greater than the current threshold, and the partial discharge amount is greater than the discharge amount threshold, and the voltage parameter is less than the voltage threshold, then the state of the power equipment is determined to be a short-circuit fault of the equipment winding; If the partial discharge amount is greater than the discharge amount threshold, and the voltage parameter is less than the voltage threshold, determining that the power equipment state is an equipment insulation aging fault; If the power parameter is greater than the power threshold, the current parameter is greater than the current threshold, and the partial discharge amount is greater than the discharge amount threshold, then the state of the power equipment is determined to be an equipment winding overheating fault.

3. The method for real-time identification of power equipment status based on artificial intelligence according to claim 1 is characterized in that: The electric power equipment feature recognition model comprises an input layer, a plurality of intermediate hidden layers and an output layer, the number of nodes in the input layer is the same as the dimension of the target three-dimensional model feature, and the activation functions of the plurality of intermediate hidden layers are different; Inputting the target three-dimensional model features into the electric power equipment feature recognition model to obtain the feature recognition results output by the electric power equipment feature recognition model includes: Inputting the target three-dimensional model features into the input layer of the electric power equipment feature recognition model, processing the target three-dimensional model features based on the first intermediate hidden layer, and outputting a first intermediate result; Processing the first intermediate result based on the second intermediate hidden layer, and outputting a second intermediate result; Processing the second intermediate result based on the third intermediate hidden layer, and outputting a third intermediate result; Processing the third intermediate result based on the output layer to output a first feature recognition result and a second feature recognition result; The first feature recognition result represents a predicted probability that the target power device is deformed, and the second feature recognition result represents a predicted probability that a defect exists at a connection of the target power device.

4. The method for real-time identification of power equipment status based on artificial intelligence according to claim 3 is characterized in that: The processing of the first intermediate hidden layer is as follows: Wherein, x1 represents the first intermediate result, x0 represents the vector corresponding to the target three-dimensional model feature; W1 represents the weight matrix of the first intermediate hidden layer, with a dimension of n*m1, where n is the dimension of the target three-dimensional model feature, and m1 is the number of nodes in the first intermediate hidden layer; b1 represents the bias vector of the first intermediate hidden layer, with a dimension of m1; Relu1(x) represents the activation function of the first intermediate hidden layer; The processing of the second intermediate hidden layer is as follows: Wherein, x2 represents the second intermediate result; W2 represents the weight matrix of the second intermediate hidden layer, the dimension is m1*m2, and m2 is the number of nodes in the second intermediate hidden layer; b2 represents the bias vector of the second intermediate hidden layer, the dimension is m2; Relu2(x) represents the activation function of the second intermediate hidden layer; The processing process of the third intermediate hidden layer is as follows: Wherein, x3 represents the third intermediate result; W3 represents the weight matrix of the third intermediate hidden layer, with a dimension of m2*m3, where m3 is the number of nodes in the third intermediate hidden layer; b3 represents the bias vector of the third intermediate hidden layer, with a dimension of m3; Relu3(x) represents the activation function of the second intermediate hidden layer; The processing of the output layer is as follows: y represents the feature recognition result, with a dimension of 2; W4 represents the weight matrix of the output layer, with a dimension of m3*2; b4 represents the bias vector of the output layer, with a dimension of 2; Relu4(x) represents the activation function of the output layer.

5. The method for real-time identification of power equipment status based on artificial intelligence according to any one of claims 1 to 4, characterized in that: The step of performing point cloud optimization on the original point cloud data to obtain target point cloud data includes: Filtering the interference point cloud in the original point cloud data to obtain filtered point cloud data; The point cloud in the filtered point cloud data is processed to enhance the information feature expression capability to obtain the target point cloud data.

6. The method for real-time identification of power equipment status based on artificial intelligence according to claim 5 is characterized in that: The filtering of the interference point cloud in the original point cloud data to obtain filtered point cloud data includes: For each first target point cloud in the original point cloud data, determine a neighborhood of each first target point cloud and a convex hull volume of the neighborhood of each first target point cloud; determining an angular discreteness of each first target point cloud based on a vector between each first target point cloud and each second target point cloud in its immediate neighborhood; Determining a spherical volume of each first target point cloud based on a distance between each first target point cloud and each second target point cloud in its immediate neighborhood; The original point cloud data is filtered based on the convex hull volume of the neighborhood of each first target point cloud, and the angular discreteness and sphere volume of each first target point cloud to obtain the filtered point cloud data.

7. The method for real-time identification of power equipment status based on artificial intelligence according to claim 5 is characterized in that: The step of performing information feature expression capability enhancement processing on the point cloud in the filtered point cloud data to obtain the target point cloud data includes: According to the spatial coordinates of each point cloud in the filtered point cloud data, the point clouds in the filtered point cloud data are grouped to obtain a plurality of point cloud subsets; Calculate the feature vector between each point cloud subset; Determine the correlation features between the point cloud subsets based on the feature vectors of each point cloud subset; New attribute information is assigned to the point cloud in each point cloud subset based on the association features between the point cloud subsets to obtain the target point cloud data.

8. A real-time identification system for power equipment status based on artificial intelligence, characterized in that: Applied to the method for real-time identification of power equipment status based on artificial intelligence according to any one of claims 1 to 7, the real-time identification system for power equipment status based on artificial intelligence comprises: A point cloud acquisition module, used for acquiring original point cloud data of a target power device in real time based on a laser radar scanning device; the target power device is any power device in the power system; A point cloud processing module, used to perform point cloud optimization on the original point cloud data to obtain target point cloud data, and perform three-dimensional reconstruction based on the target point cloud data to obtain target three-dimensional model features of the target power equipment; A feature recognition module, used for inputting the target three-dimensional model features into the power equipment feature recognition model to obtain the feature recognition results output by the power equipment feature recognition model; the power equipment feature recognition model is trained based on the sample three-dimensional model features and their corresponding result labels; The power equipment state identification module is used to determine the power equipment state of the target power equipment based on the feature identification result.

9. An electronic device, comprising: Memory for storing computer software programs; A processor, used to read and execute the computer software program, characterized in that when the processor executes the computer software program, it implements the real-time identification method of the power equipment status based on artificial intelligence as described in any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium having a computer software program stored therein, characterized in that: When the computer software program is executed by the processor, the method for real-time identification of the state of electric power equipment based on artificial intelligence as claimed in any one of claims 1 to 7 is implemented.

Images