An automatic inspection method for an overhead transmission line and related products

Abstract

The present application discloses an automatic inspection method and related products for overhead transmission lines, which can be applied to the technical field of circuit inspection. The method includes: obtaining the to-be-classified point cloud data corresponding to the area to be inspected; there are ground objects and power lines in the area to be inspected; using a lightweight transmission line point cloud segmentation network to classify the to-be-classified point cloud data, and obtaining a first point cloud data set corresponding to the ground objects and a second point cloud data set corresponding to the power lines; processing the first point cloud data set based on the point cloud thinning algorithm, and obtaining a first comparison data set; determining the central axis point set data of the power lines based on the second point cloud data set; calculating the distances between the points in the first comparison data set and the points in the central axis point set data, and comparing with a preset safety distance to determine the inspection result. In this way, the present application introduces an FPS network that can quickly classify, and reduces the point cloud density through the point cloud thinning algorithm and the concept of the central axis to improve the inspection efficiency.

Description

Technical Field

[0001] This application relates to the technical field of circuit inspection, and particularly to an automatic inspection method and related products for overhead transmission lines. Background Art

[0002] An overhead transmission line consists of electric towers and power lines. The electric towers support the power lines in the air and conduct power transmission. As an important part of the power transmission system, overhead transmission lines have many safety management rules, such as strictly prohibiting the stacking of items below. Therefore, regular inspection of overhead transmission lines is crucial.

[0003] Existing inspection methods for overhead transmission lines mostly rely on manual inspection by technical personnel on-site. However, for overhead transmission lines with a wide range and different and complex on-site environments in different regions, manual inspection will lead to low inspection efficiency.

[0004] Therefore, how to improve the inspection efficiency is an urgent problem to be solved by those skilled in the art. Summary of the Invention

[0005] Based on the above problems, this application provides an automatic inspection method and related products for overhead transmission lines, introducing an FPS network that can quickly classify, and reducing the point cloud density through a point cloud thinning algorithm and the concept of the central axis, thereby improving the inspection efficiency.

[0006] In a first aspect, an embodiment of this application provides an automatic inspection method for overhead transmission lines, including:

[0007] Obtain the to-be-classified point cloud data corresponding to the area to be inspected; there are ground objects and power lines in the area to be inspected;

[0008] Use a lightweight transmission line point cloud segmentation network to classify the to-be-classified point cloud data, and obtain a first point cloud data set corresponding to the ground objects and a second point cloud data set corresponding to the power lines;

[0009] Process the first point cloud data set based on the point cloud thinning algorithm, and obtain a first comparison data set;

[0010] Determine the central axis point set data of the power lines based on the second point cloud data set;

[0011] Calculate the distances between the points in the first comparison data set and the points in the central axis point set data, and compare them with a preset safety distance to determine the inspection result.

[0012] Optionally, the obtaining the to-be-classified point cloud data corresponding to the area to be inspected includes:

[0013] Obtain the point cloud data to be processed corresponding to the area to be inspected;

[0014] Based on the point cloud completion algorithm, complete the point cloud data to be processed and obtain the point cloud data to be classified.

[0015] Optionally, the step of using the point cloud completion algorithm to complete the point cloud data to be processed and obtain the point cloud data to be classified includes:

[0016] Based on the linear interpolation algorithm and the quadratic curve model, complete the missing point cloud data corresponding to the power line and obtain the point cloud data to be classified.

[0017] Optionally, the step of processing the first point cloud data set based on the point cloud thinning algorithm and obtaining the first comparison data set includes:

[0018] Divide the first point cloud data set into a preset number of network blocks with the same spatial size;

[0019] Determine the central point positions corresponding to each network block;

[0020] Among all the network blocks, determine the network blocks where the distance between the central point position and the power line exceeds the first preset distance as the target network blocks;

[0021] Randomly delete half of the point cloud data corresponding to the target network blocks, and integrate the remaining point cloud data to obtain the first comparison data set.

[0022] Optionally, the step of determining the central axis point set data of the power line based on the second point cloud data set includes:

[0023] Based on the spline curve interpolation algorithm, transform the second point cloud data set and obtain a dense and smooth point cloud data;

[0024] Based on the smooth point cloud data, determine the central axis point set data of the power line.

[0025] Optionally, the method further includes:

[0026] Evaluate the inspection result based on the fuzzy comprehensive evaluation method and obtain the comprehensive evaluation result;

[0027] Compare the comprehensive evaluation result with the preset evaluation table to determine whether the clearance of the area to be inspected is in a safe state.

[0028] Optionally, the step of evaluating the inspection result based on the fuzzy comprehensive evaluation method and obtaining the comprehensive evaluation result includes:

[0029] Determine the influencing factors of clearance safety based on the inspection result;

[0030] Calculate the membership degree of the factors affecting the clearance safety and obtain the calculation results;

[0031] Normalize the calculation results to obtain the comprehensive evaluation results.

[0032] In a second aspect, an embodiment of the present application provides an automatic inspection device for an overhead transmission line, including:

[0033] An acquisition module for acquiring the point cloud data to be classified corresponding to the area to be inspected; there are ground objects and power lines in the area to be inspected;

[0034] A classification module for classifying the point cloud data to be classified by using a lightweight transmission line point cloud segmentation network, and obtaining a first point cloud data set corresponding to the ground objects and a second point cloud data set corresponding to the power lines;

[0035] A first processing module for processing the first point cloud data set based on a point cloud thinning algorithm and obtaining a first comparison data set;

[0036] A second processing module for determining the central axis point set data of the power line based on the second point cloud data set;

[0037] A detection module for calculating the distances between the points in the first comparison data set and the points in the central axis point set data, and comparing with a preset safety distance to determine the inspection result.

[0038] In a third aspect, an embodiment of the present application provides an automatic inspection device for an overhead transmission line, including:

[0039] A memory for storing a computer program;

[0040] A processor for implementing the steps of the automatic inspection method for an overhead transmission line as described above when executing the computer program.

[0041] In a fourth aspect, an embodiment of the present application provides a readable storage medium, on which a computer program is stored, and the computer program realizes the steps of the automatic inspection method for an overhead transmission line as described above when executed by a processor.

[0042] It can be seen from the above technical solutions that compared with the prior art, the present application has the following advantages:

[0043] This application first obtains the point cloud data to be classified corresponding to the area to be inspected. Among them, there are ground objects and power lines in the area to be inspected. Then, a lightweight transmission line point cloud segmentation network is used to classify the point cloud data to be classified, and a first point cloud data set corresponding to the ground objects and a second point cloud data set corresponding to the power lines are obtained. Finally, the first point cloud data set is processed based on the point cloud thinning algorithm to obtain a first comparison data set. The central axis point set data of the power line is determined based on the second point cloud data set, and the distances between the points in the first comparison data set and the points in the central axis point set data are calculated and compared with the preset safety distance to determine the inspection result. In this way, the embodiment of this application introduces an FPS network that can quickly classify, and reduces the point cloud density through the point cloud thinning algorithm and the concept of the central axis, improving the inspection efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 It is a flowchart of an automatic inspection method for an overhead transmission line provided by an embodiment of this application;

[0045] Figure 2 It is a schematic diagram of the architecture of an FPS network for rapid segmentation of an overhead transmission line point cloud scene provided by an embodiment of this application;

[0046] Figure 3 It is a schematic diagram of the structure of an automatic inspection device for an overhead transmission line provided by an embodiment of this application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0047] As described above, the existing inspection methods for overhead transmission lines have the problem of low inspection efficiency. Specifically, the existing inspection methods for overhead transmission lines mostly rely on manual inspection by technical personnel on-site. However, for overhead transmission lines with a wide span and different and complex on-site environments in different regions, in order to ensure the inspection quality, technical personnel need to carefully inspect the overhead transmission lines in each area, resulting in the problem of low inspection efficiency. In addition, the prior art also uses unmanned aerial vehicles to inspect overhead transmission lines. Compared with manual inspection, unmanned aerial vehicles can reach some places that people cannot reach, and the data collected is also huge. However, in the face of a huge amount of collected data, the prior art usually uses a full-scale comparison method to determine the inspection result, which also leads to the problem of low inspection efficiency.

[0048] To solve the above problems, an embodiment of the present application provides an automatic inspection method for overhead transmission lines, including: First, obtain the to-be-classified point cloud data corresponding to the area to be inspected. Among them, there are ground objects and power lines in the area to be inspected. Then, use a lightweight transmission line point cloud segmentation network to classify the to-be-classified point cloud data, and obtain a first point cloud data set corresponding to the ground objects and a second point cloud data set corresponding to the power lines. Finally, process the first point cloud data set based on the point cloud thinning algorithm, and obtain a first comparison data set. Determine the central axis point set data of the power line based on the second point cloud data set, calculate the distances between the points in the first comparison data set and the points in the central axis point set data, compare with the preset safety distance, and determine the inspection result.

[0049] In this way, the embodiment of the present application introduces an FPS network that can quickly classify, and reduces the point cloud density through the point cloud thinning algorithm and the concept of the central axis, improving the inspection efficiency.

[0050] It should be noted that an automatic inspection method and related products for overhead transmission lines provided by the present application can be applied to the field of circuit inspection technology. The above is only an example and does not limit the application field of an automatic inspection method and related products for overhead transmission lines provided by the present application.

[0051] In order to make the objectives, technical solutions, and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present application.

[0052] Figure 1 It is a flowchart of an automatic inspection method for overhead transmission lines provided by an embodiment of the present application. Combined with Figure 1 As shown, an automatic inspection method for overhead transmission lines provided by an embodiment of the present application may include:

[0053] S101: Obtain the to-be-classified point cloud data corresponding to the area to be inspected; there are ground objects and power lines in the area to be inspected.

[0054] In practical applications, the rapid inspection of transmission lines is the basic guarantee for ensuring the safety of the line network. With the increasing pressure of line inspection brought about by the growing transmission network and in order to better ensure the safety of inspection personnel, unmanned aerial vehicle (UAV) inspection technology has been more widely applied. In the embodiments of this application, UAVs are used for data collection. After the UAVs upload the collected data to the system, the server calculates the inspection results based on the collected data. Specifically, the UAVs are equipped with imaging equipment and can operate according to the routes formulated by technical personnel and collect data through the imaging devices. The positions of each electric tower and the mounting forms of the power lines on the electric towers are recorded. Technical personnel can determine the area to be inspected based on the records and formulate inspection routes for the UAVs for this area. It should be noted that in order to ensure the effectiveness of the inspection and determine the clearance safety level of the power lines, there must be ground features and power lines in the area to be inspected. When the UAVs fly on the inspection routes, they can collect the point cloud data (point cloud data to be classified) in the area to be inspected through the mounted imaging devices and upload it. In addition, since the power lines are continuous, it is more appropriate to use the electric towers as markers to distinguish geographical locations. Therefore, when dividing the area to be inspected, the area between two electric towers is usually used as an area to be inspected. It can be understood that the area to be inspected usually also includes the electric towers. The point cloud data to be classified corresponding to the area to be inspected obtained by the UAVs includes the point cloud data of ground features, the point cloud data of power lines, the point cloud data of electric towers, and the point cloud data of other categories (such as flying birds, etc.).

[0055] In addition, since the methods for obtaining the point cloud data to be classified are not the same, the embodiments of this application can illustrate one possible obtaining method.

[0056] In one case, obtaining the point cloud data to be classified corresponding to the area to be inspected includes:

[0057] Obtaining the point cloud data to be processed corresponding to the area to be inspected;

[0058] Completing the point cloud data to be processed based on a point cloud completion algorithm and obtaining the point cloud data to be classified.

[0059] In practical applications, during the process of the UAVs scanning the point cloud of the area to be inspected through lidar or imaging devices, they will be interfered by factors such as weather, environment, and electromagnetic signals, resulting in the problem of missing point cloud data. Therefore, the embodiments of this application need to complete the missing point cloud data. Specifically, the embodiments of this application assume that the point cloud data collected by the UAVs is missing. After the UAVs upload this point cloud data (point cloud data to be processed), the server needs to use a point cloud completion algorithm centered on the application of quadratic functions and symmetry principles to complete the processing of this point cloud data to be processed, and then obtain the point cloud data to be classified.

[0060] In addition, since the methods for completing the to-be-processed point cloud data are different, the embodiments of the present application may illustrate one possible completion method.

[0061] In one case, completing the to-be-processed point cloud data based on the point cloud completion algorithm and obtaining the to-be-classified point cloud data includes:

[0062] Completing the missing point cloud data corresponding to the power line based on the linear interpolation algorithm and the quadratic curve model, and obtaining the to-be-classified point cloud data.

[0063] In practical applications, taking the case where there are ground objects, power lines, and electric towers in the area to be inspected as an example, the point cloud data that needs to be supplemented is the missing point cloud data of the power lines and electric towers. Therefore, the server needs to first perform a preliminary classification on the acquired to-be-processed point cloud data to determine the point cloud data corresponding to the power lines and the point cloud data corresponding to the electric towers. The embodiments of the present application provide a lightweight point cloud (Fast Powerline Segmentation, FPS) network for fast segmentation in the overhead transmission line scenario, which can complete the classification of point cloud data in a relatively short time. For the completion of the point cloud data corresponding to the power lines, first calculate the distance between adjacent power line points. If it exceeds the set threshold, it is marked as a break point. It can be understood that assuming there are multiple parallel power lines between two electric towers, then a coordinate system is established with the power lines parallel to the X-axis. The distance between two points with the same X coordinate on two adjacent power lines should be within a certain range. Therefore, a threshold can be set. If the distance between two points exceeds the set threshold, it is marked as a break point. Then, connect the two points near the break point into a straight line, and fit a straight line according to the two-point coordinates, as shown in the following formula (1):

[0064]

[0065] where (X 1 , Y 1 , Z 1 ) and (X 2 , Y 2 , Z 2 ) are two known adjacent break points, t represents the parameter of the three-dimensional straight line equation, and (x, y, z) represents the coordinate points generated near the break point. Finally, in the three-dimensional coordinates (X 1 , Y 1 , Z 1 ) and (X 2 , Y 2 , Z 2) Take equally spaced values between them and perform interpolation calculations to obtain a total of i coordinate points, that is, obtain i interpolation points between the break points, achieving rapid completion of the locally missing power lines. According to this method, the point cloud data of the power lines in the area to be inspected is completed, and the complete power line point cloud data is obtained. For the completion of the point cloud data corresponding to the electric tower, it is based on the principle of symmetry on both sides of the electric tower to complete the locally missing point cloud data of the electric tower. Specifically, first read the three-dimensional data of the electric tower point cloud, and then divide it into several parts according to the average height difference by calculating the height of the point cloud. Taking the example of dividing it into 100 equal parts, count the number of points in each height interval and calculate the center point (x mean ,y mean ,z mean ) of each interval. The calculation formula is as follows:

[0066]

[0067] In the formula, n represents the total number of point clouds in this interval, and j represents the serial number of the point clouds in the interval. When divided equally by height, the center point of the set calculated for the interval with missing point clouds may deviate from the true position. Therefore, it is necessary to use negative feedback to fine-tune it so that the center of the missing point clouds gradually returns to the true position. Specifically, first set the slope and intercept of the initial straight line in combination with the center points of the upper and lower adjacent partitions, and define the maximum number of iterations. During the iteration process, continuously fit the straight line and calculate the fitting error of each point. According to the fitting error, perform negative feedback adjustment on the slope and intercept. If the error is greater than 0, reduce the values of the slope and intercept to make the fitting straight line result closer to the overall center point set globally, and finally determine the symmetry line of the electric tower. Then refer to the symmetry line to complete the missing point cloud of the local electric tower. Assume that the coordinates of the two symmetric points are (x j , y j , z j ) and (x’ j , y’ j , z’ j ), and the two point coordinates satisfy the following formula:

[0068] 2*x mean =x j +x′ j (5)

[0069] 2*y mean =y j +y′ j (6)

[0070] z j =z′ j (7)

[0071] Then the coordinates of the symmetric point are (x’ j , y’ j, z’ j ) can be determined as:

[0072] x’ j = 2 * x mean - x j (8)

[0073] y’ j = 2 * y mean - y j (9)

[0074] z’ j = z j (10)

[0075] Based on this, the point cloud data with local missing parts of the electric tower is completed, and the complete point cloud data of the electric tower is obtained. The complete point cloud data of the power line, the complete point cloud data of the electric tower, and the point cloud data of the ground objects are integrated to obtain the point cloud data to be classified.

[0076] S102: Classify the point cloud data to be classified by using a lightweight transmission line point cloud segmentation network, and obtain a first point cloud data set corresponding to the ground objects and a second point cloud data set corresponding to the power line.

[0077] In practical applications, considering the real-time performance of the inspection task and the transfer and generalization ability of the deep learning network, in order to improve the inspection efficiency, the embodiments of the present application have carried out multiple speed tests, adjusted the number of network nodes, and designed an FPS network specifically for rapid segmentation of the overhead transmission line point cloud scene. Figure 2 It is a schematic diagram of the architecture of an FPS network for rapid segmentation of an overhead transmission line point cloud scene provided by the embodiments of the present application. Combining Figure 2As shown in the figure, the network extracts features and classifies the input point cloud data to be classified through multiple fully connected layers. (N,6)-xyz&rgb indicates that the network input includes 6-dimensional point cloud data to be classified, including 3-dimensional spatial coordinates and 3-dimensional color information. The output of the network is the probability of each point cloud data corresponding to each category. Specifically, according to the main concerns of power line inspection, three categories are set for the network, including power lines, ground objects, and electric towers. The input layer receives 6-dimensional point cloud data to be classified. The original coordinate system is converted into the local coordinate system where a single test set is located through a decentralization operation, further highlighting the height difference between the wire, tower, and ground, facilitating the network to learn more features. First, the first fully connected layer (FC1) maps the input data to a 64-dimensional feature space, and the ReLU activation function performs a non-linear transformation on the output of the first layer. Then, the second fully connected layer (FC2) maps the output of the first layer to a 64-dimensional feature space, and the ReLU activation function performs a non-linear transformation on the output of the second layer. Finally, the third fully connected layer (FC3) maps the output of the second layer to the probability distribution of three categories, and the output layer outputs the probability that the point cloud data to be classified belongs to each category. It is set for the network that when the probability exceeds the preset threshold, it is determined that the point cloud data to be classified belongs to a certain category. In this way, the FPS network for rapid segmentation of the overhead transmission line point cloud scene quickly classifies the point cloud data to be classified, classifies the point cloud data corresponding to the ground objects into the first point cloud data set, classifies the point cloud data corresponding to the power lines into the second point cloud data set, and classifies the point cloud data corresponding to the electric towers into the third point cloud data set. In addition, for the training of the FPS network for rapid segmentation of the overhead transmission line point cloud scene, the embodiments of the present application provide a training method. Specifically, first, a large amount of training data (point cloud data of power lines, point cloud data of electric towers, and point cloud data of ground objects) is provided, and the training data is divided into small batches for training. The number of each batch and the number of training repetitions are set by technicians themselves, such as 32 for each batch and 500 epochs for each training repetition, etc. The set data loading function in the code is used to receive the point cloud data and convert the data into tensor form. The set custom dataset function is used to encapsulate the dataset and define the length of the dataset and the data acquisition method. The set point cloud classification model function defines the structure of the network, including three fully connected layers and the ReLU activation function. The training function is used to train the model, save the model parameters with better results, use the cross-entropy loss function to measure the difference between the network output result and the true label, and use the Adam optimizer to update the network parameters, thereby realizing the training of the network and obtaining the FPS network for rapid segmentation of the overhead transmission line point cloud scene for the embodiments of the present application.

[0078] S103: Process the first point cloud data set based on the point cloud thinning algorithm and obtain the first comparison data set.

[0079] In practical applications, the FPS network for rapid segmentation of the overhead transmission line point cloud scene is used to classify the point cloud data to be classified, obtaining the first point cloud data set corresponding to the ground objects, the second point cloud data set corresponding to the power lines, and the third point cloud data set corresponding to the electric towers. It can be understood that the point cloud data in the data set is huge, and if calculations are based on this, the calculation process will be relatively long, thus reducing the inspection efficiency. For this reason, the embodiments of the present application provide a method for reducing the point cloud density to reduce the amount of calculation and improve the inspection efficiency. Specifically, for the first point cloud data set corresponding to the ground objects, the embodiments of the present application process it based on the point cloud thinning algorithm. Since the present application mainly considers the clearance safety level of the overhead transmission line, the closer the ground object is to the power line, the greater the impact. The server can divide the ground objects in terms of height and set a threshold, taking the ground objects exceeding a certain height as the key calculation targets without processing; taking the ground objects below a certain height as non-key calculation targets and performing point cloud thinning processing.

[0080] In addition, since the methods for processing the first point cloud data set are not the same, the embodiments of the present application can illustrate a possible processing method.

[0081] In one case, S103: Process the first point cloud data set based on the point cloud thinning algorithm and obtain a first comparison data set, which may specifically include:

[0082] Divide a preset number of network blocks with the same spatial size based on the first point cloud data set;

[0083] Determine the central point positions corresponding to each network block;

[0084] Among all the network blocks, determine the network blocks whose distance between the central point position and the power line exceeds a first preset distance as target network blocks;

[0085] Randomly delete half of the point cloud data corresponding to the target network blocks, and integrate the remaining point cloud data to obtain the first comparison data set.

[0086] In practical applications, when the server thins the point cloud, it can evenly divide the spatial position where the first point cloud data set is located into several grid blocks of the same size. The number of grid blocks can be 500 or other numbers. It can be understood that the divided grid blocks are small cuboids stacked together, and these small cuboids are stacked together to form a large cuboid, whose upper surface is flush with the lower surface of the power line, and the lower surface is flush with the lowest point in the first point cloud data set. Taking the number of grid blocks as 500 as an example, define the divide_into_blocks(data_set, num_blocks = 500) function. First, calculate the specification size of the first point cloud data set, and use list comprehension to divide it into blocks. Then traverse the corresponding point cloud positions in each grid block and calculate the center point position. Calculate the average value in each direction through the np.mean function to determine the center point of the grid block. Using axis = 0 means calculating the mean value column by column to perform the averaging operation on each column, and there is the following expression:

[0087]

[0088] Then traverse and calculate the Euclidean distance from this center point to each point in the second data set of the data set, and store it in a table. Finally, return the minimum distance in the list. The expression is as follows:

[0089]

[0090] In summary, in the formula, P is the set of coordinate points in the second data set; axis represents the point cloud serial number of the center point; m represents the number of point clouds; q i represents the point cloud serial number in the grid area; p k represents the point cloud serial number of the second data set corresponding to the power line. The technical personnel set a threshold (usually 30 meters). For grid blocks with a distance greater than the threshold from the power line, randomly delete half of the data points from this block. In this way, integrate the remaining point cloud data to form a new data set for comparison, that is, the first comparison data set.

[0091] S104: Determine the data set of the central axis points of the power line based on the second point cloud data set.

[0092] In practical applications, multiple power lines are hung between two power towers, and the second point cloud dataset includes all the point cloud data of these power lines. It can be understood that this data is huge, and if calculations are based on this, the calculation process will be relatively long, thus reducing the inspection efficiency. Therefore, the embodiments of the present application provide a method for reducing the point cloud density to reduce the amount of calculation and improve the inspection efficiency. Specifically, for the second point cloud dataset corresponding to the power line, the embodiments of the present application fit multiple strands of power lines between adjacent power towers into a power line central axis with similar curvature, and calculate based on the point set data of the central axis of the power line and the above first comparison dataset, so as to achieve the purpose of reducing the point cloud density and improving the inspection efficiency.

[0093] In addition, since the methods for processing the second point cloud dataset are not the same, the embodiments of the present application can illustrate one possible processing method.

[0094] In one case, S104: Determining the point set data of the central axis of the power line based on the second point cloud dataset may specifically include:

[0095] Converting the second point cloud dataset based on the spline curve interpolation algorithm to obtain dense and smooth point cloud data;

[0096] Determining the point set data of the central axis of the power line based on the smooth point cloud data.

[0097] In practical applications, multiple power lines are connected between power towers. Clearance detection needs to ensure that there are no potential ground object hazards within the clearance range of each power line. However, if each power line is detected, it will inevitably affect the operation efficiency. To improve the operation efficiency, several power lines between two power towers are reasonably fitted into a power line central axis. Through the combination of interpolation and coordinate calculation, the point cloud data of the power line is converted into the point set data of the central axis of the power line. Specifically, the server first calculates the average value of the coordinates of every two points in the second point cloud dataset corresponding to the power line, and then obtains the midpoint. The point cloud data in the second point cloud dataset is converted into denser and smoother point cloud data through spline curve interpolation. The number of spline interpolations, smoothness, etc. are set by technicians. Generally, the number of interpolation times k = 3, and the smoothness s is 0. Set a t, which is used to represent a numerical sequence generated from 0 to 1, and the number of elements in the sequence is the same as the number of discrete points of the power line (the number of coordinate point cloud data in the second point cloud dataset), so as to calculate the points on the central axis. Then, calculate the index position of the linear interpolation according to t. Specifically, multiply t by the number of points of the discrete curve and take the integer part. From the index sequence of the corresponding curve coordinates, calculate the central axis (x, y) coordinates of the points at the corresponding index positions on each curve according to the parameter t, as shown in the following expression (13):

[0098] n(x,y) = [t·num] (13)

[0099] Where n(x,y) is the index sequence corresponding to the curve coordinates, and num is the number of points on the discrete curve. Finally, by averaging these points, virtual points on the central axis of the power line at each position are obtained. The parameter positions are adjusted by linear interpolation to make the virtual points more consistent with the sag of the actual wire points, realizing the fitting of multi-strand power lines, and obtaining the central axis point set data as the data set for subsequent calculation with the first comparison data set.

[0100] S105: Calculate the distances between the points in the first comparison data set and the points in the central axis point set data, and compare them with a preset safety distance to determine the inspection result.

[0101] In practical applications, regarding the detection of the clearance in the vicinity of the transmission line, the safety distance for 500 kV in the operating regulations of overhead transmission lines is 8.5 m. The distances from the central axis of the general power line to the single-strand power line are 18 m and 12 m respectively. Therefore, when the server uses the power line central axis as the reference for clearance inspection, the safety distance can be appropriately adjusted, and the recommended value is between 8.5 m and 26.5 m. Thus, when setting 10 m as the preset safety distance, the server can first draw a circle perpendicular to the power line with each point on the central axis as the center, and then form a cylinder with a radius of 10 m from these circles. Then, traverse each point cloud data in the first comparison data set, calculate the distances between its points and the points in the central axis point set data, realize the retrieval of the space in the vicinity of the central axis. When the distance from a point in the first comparison data set to a point in the central axis point set data is less than or equal to 10 m, the inspection result within the safe range around the power line for this point is obtained. That is, it is detected that there are ground objects around the power line. When the distance does not meet the safety requirements in the specification, the algorithm automatically alarms, as shown in formula (14):

[0102]

[0103] Where P′ mid (x′ mid ,y′ mid ,z′ mid ) represents the points on the central axis of the power line, f represents the points of the surrounding ground object environment, and r represents the neighborhood search radius.

[0104] In addition, since the processing methods for the inspection results are not the same, the embodiments of the present application can illustrate one possible processing method.

[0105] In one case, the method further includes:

[0106] Evaluating the inspection result based on the fuzzy comprehensive evaluation method and obtaining a comprehensive evaluation result;

[0107] The comprehensive evaluation result is compared with a preset evaluation table to determine whether the clearance of the area to be inspected is in a safe state.

[0108] In practical applications, whether the ground object is within the safety range of the power line is the inspection result obtained by the above method. The embodiment of the present application also provides an overhead power transmission line clearance safety evaluation system based on the fuzzy comprehensive evaluation method, and further clearance safety evaluation is triggered when the ground object is within the safety range of the power line. Specifically, there are many types of ground objects, such as trees, three-span buildings, and surrounding scene power grids. Different types have different degrees of influence on the clearance safety of transmission lines. Therefore, the above inspection results can be further evaluated by the fuzzy comprehensive evaluation method, and a numerical comprehensive evaluation result can be obtained, such as 1, 2, and 10. The preset evaluation table is used to indicate the clearance safety situation corresponding to different values. Taking levels 1 to 3 as an example, the higher the level, the lower the safety level of the circuit line, and each level has its corresponding comprehensive evaluation result numerical range. Compare the comprehensive evaluation result with the preset evaluation table. If the value corresponding to level 3 is 8-10, and the value of the comprehensive evaluation result is determined by calculation to be 9.1, the clearance safety status of the area to be inspected determined and returned by the server is level 3. If level 3 is predefined as an unsafe state, the server will also warn that the inspection area is clear and in an unsafe state.

[0109] In addition, since the methods for evaluating the inspection results are not the same, the embodiments of the present application can illustrate a possible evaluation method.

[0110] In one case, the inspection result is evaluated based on the fuzzy comprehensive evaluation method to obtain a comprehensive evaluation result, including:

[0111] Determining clearance safety influencing factors based on the inspection results;

[0112] Calculating the membership degree of the clearance safety influencing factors and obtaining the calculation results;

[0113] The calculation results are normalized to obtain the comprehensive evaluation results.

[0114] In practical applications, the overhead transmission line clearance safety evaluation system based on the fuzzy comprehensive evaluation method mainly includes the following steps. Specifically, first, determine the clearance safety influencing factors, and input the tree growth height, three-span buildings, and the closest distance to the surrounding scene power grid in this area. Then, establish a fuzzy evaluation matrix (weight vector matrix A) for each safety influencing factor. Among them, the elements in the matrix represent the membership degrees of calcium elements at different levels. The determination of the membership degrees can be obtained through expert opinions, on-site investigations, or historical data, etc., and the sum of the weight vectors needs to be equal to 1. Then, obtain the evaluation situations corresponding to each index. During the evaluation process, combine the above various algorithms to analyze the results and construct a weight judgment matrix R, and perform membership degree calculation (matrix cross multiplication) on matrix A and matrix R to obtain the calculation results of each scheme. Finally, perform normalization processing according to the membership degree calculation results, compare the obtained comprehensive score with the evaluation form to obtain the comprehensive evaluation result, and determine whether the state of the power grid clearance in this area is safe.

[0115] In summary, this application first obtains the to-be-classified point cloud data corresponding to the area to be inspected. Among them, there are ground objects and power lines in the area to be inspected. Then, use the lightweight transmission line point cloud segmentation network to classify the to-be-classified point cloud data, and obtain the first point cloud data set corresponding to the ground objects and the second point cloud data set corresponding to the power lines. Finally, process the first point cloud data set based on the point cloud thinning algorithm to obtain the first comparison data set, determine the central axis point set data of the power lines based on the second point cloud data set, calculate the distances between the points in the first comparison data set and the points in the central axis point set data, compare with the preset safety distance, and determine the inspection result. In this way, the embodiment of this application introduces an FPS network that can quickly classify, and reduces the point cloud density through the point cloud thinning algorithm and the concept of the central axis, improving the inspection efficiency.

[0116] Based on the automatic inspection method for overhead transmission lines provided in the above embodiments, the embodiment of this application also provides an automatic inspection device for overhead transmission lines. Next, the automatic inspection device for overhead transmission lines will be described in combination with the embodiments and the drawings respectively.

[0117] Figure 3 It is a schematic structural diagram of an automatic inspection device for overhead transmission lines provided by an embodiment of this application. Combining Figure 3 As shown, the automatic inspection device 300 for overhead transmission lines provided by the embodiment of this application includes:

[0118] An acquisition module 301, configured to acquire the to-be-classified point cloud data corresponding to the area to be inspected; there are ground objects and power lines in the area to be inspected;

[0119] The classification module 302 is used to classify the point cloud data to be classified by using a lightweight transmission line point cloud segmentation network, and obtain a first point cloud data set corresponding to the ground object and a second point cloud data set corresponding to the power line;

[0120] The first processing module 303 is used to process the first point cloud data set based on a point cloud thinning algorithm and obtain a first comparison data set;

[0121] The second processing module 304 is used to determine the central axis point set data of the power line based on the second point cloud data set;

[0122] The detection module 305 is used to calculate the distances between the points in the first comparison data set and the points in the central axis point set data, compare them with a preset safety distance, and determine the inspection result.

[0123] As an implementation manner, for how to obtain the point cloud data to be classified corresponding to the area to be inspected, the above-mentioned acquisition module 301 includes: an acquisition sub-module and a completion module;

[0124] The acquisition sub-module is used to acquire the point cloud data to be processed corresponding to the area to be inspected;

[0125] The completion module is used to complete the point cloud data to be processed based on a point cloud completion algorithm and obtain the point cloud data to be classified.

[0126] In the first implementation manner, for how to complete the point cloud data to be processed, the above-mentioned completion module is specifically used for:

[0127] Completing the missing point cloud data corresponding to the power line based on a linear interpolation algorithm and a quadratic curve model, and obtaining the point cloud data to be classified.

[0128] As an implementation manner, for how to process the first point cloud data set, the above-mentioned first processing module 303 is specifically used for:

[0129] Dividing a preset number of network blocks with the same spatial size based on the first point cloud data set;

[0130] Determining the central point positions corresponding to each network block;

[0131] Determining the network blocks whose distances between the central point positions and the power line exceed a first preset distance among all the network blocks as target network blocks;

[0132] Randomly deleting the point cloud data corresponding to half of the target network blocks, and integrating the remaining point cloud data to obtain a first comparison data set.

[0133] As an implementation manner, for how to process the second point cloud data set, the above-mentioned second processing module 304 is specifically configured to:

[0134] Convert the second point cloud data set based on the spline curve interpolation algorithm, and obtain dense and smooth point cloud data;

[0135] Determine the central axis point set data of the power line based on the smooth point cloud data.

[0136] As an implementation manner, for how to perform evaluation, the above-mentioned automatic inspection device 300 for overhead transmission lines further includes: an evaluation module;

[0137] The evaluation module is used to evaluate the inspection result based on the fuzzy comprehensive evaluation method and obtain a comprehensive evaluation result;

[0138] Compare the comprehensive evaluation result with a preset evaluation table to determine whether the clearance of the area to be inspected is in a safe state.

[0139] As an implementation manner, for how to evaluate the inspection result based on the fuzzy comprehensive evaluation method, the above-mentioned evaluation module is specifically configured to:

[0140] Determine the clearance safety influencing factors based on the inspection result;

[0141] Calculate the membership degree of the clearance safety influencing factors and obtain a calculation result;

[0142] Perform normalization processing on the calculation result to obtain a comprehensive evaluation result.

[0143] In summary, the present application first obtains the to-be-classified point cloud data corresponding to the area to be inspected. Among them, there are ground objects and power lines in the area to be inspected. Then, the lightweight transmission line point cloud segmentation network is used to classify the to-be-classified point cloud data, and the first point cloud data set corresponding to the ground object and the second point cloud data set corresponding to the power line are obtained. Finally, the first point cloud data set is processed based on the point cloud thinning algorithm, and the first comparison data set is obtained. The central axis point set data of the power line is determined based on the second point cloud data set, and the distance between each point in the first comparison data set and each point in the central axis point set data is calculated and compared with the preset safety distance to determine the inspection result. In this way, the embodiment of the present application introduces an FPS network that can quickly classify, and reduces the point cloud density through the point cloud thinning algorithm and the concept of the central axis, improving the inspection efficiency.

[0144] In addition, the embodiment of the present application also provides an automatic inspection device for overhead transmission lines, including:

[0145] A memory for storing a computer program;

[0146] A processor, which is configured to implement the steps of the automatic inspection method for an overhead transmission line as described above when executing the computer program.

[0147] In addition, an embodiment of the present application further provides a readable storage medium, on which a computer program is stored, and the computer program is configured to implement the steps of the automatic inspection method for an overhead transmission line as described above when being executed by a processor.

[0148] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An automatic inspection method for overhead power transmission lines, characterized in that: The method comprises: Acquire the point cloud data to be classified corresponding to the area to be inspected; the area to be inspected has ground objects and power lines; Classifying the point cloud data to be classified using a lightweight power transmission line point cloud segmentation network, and obtaining a first point cloud data set corresponding to the ground object and a second point cloud data set corresponding to the power line; Processing the first point cloud data set based on a point cloud thinning algorithm to obtain a first comparison data set; Determine the central axis point set data of the power line based on the second point cloud data set; Calculate the distance between each point in the first comparison data set and each point in the central axis point set data, and compare with the preset safety distance to determine the inspection result; The step of obtaining the point cloud data to be classified corresponding to the area to be inspected includes: Obtain the point cloud data to be processed corresponding to the area to be inspected; Completing the point cloud data to be processed based on a point cloud completion algorithm to obtain point cloud data to be classified; The step of processing the first point cloud data set based on a point cloud thinning algorithm to obtain a first comparison data set includes: Divide a preset number of network blocks of the same spatial size based on the first point cloud data set; Determine the center point location corresponding to each network block; Determine, from among all the network blocks, a network block whose distance from the center point to the power line exceeds a first preset distance as a target network block; Randomly delete half of the point cloud data corresponding to the target network block, and integrate the remaining point cloud data to obtain a first comparison data set.

2. The method according to claim 1, characterized in that The step of completing the point cloud data to be processed based on a point cloud completion algorithm and obtaining the point cloud data to be classified includes: Based on the linear interpolation algorithm and the quadratic curve model, the missing point cloud data corresponding to the power lines are completed, and the point cloud data to be classified is obtained.

3. The method according to claim 1, characterized in that The determining the central axis point set data of the power line based on the second point cloud data set includes: Converting the second point cloud data set based on a spline curve interpolation algorithm to obtain dense and smooth point cloud data; The central axis point set data of the power line is determined based on the smoothed point cloud data.

4. The method according to claim 1, characterized in that: The method further comprises: Evaluate the inspection results based on the fuzzy comprehensive evaluation method and obtain a comprehensive evaluation result; The comprehensive evaluation result is compared with a preset evaluation table to determine whether the clearance of the area to be inspected is in a safe state.

5. The method according to claim 4, characterized in that The inspection results are evaluated based on the fuzzy comprehensive evaluation method to obtain a comprehensive evaluation result, including: Determining clearance safety influencing factors based on the inspection results; Calculating the membership degree of the clearance safety influencing factors and obtaining the calculation results; The calculation results are normalized to obtain the comprehensive evaluation results.

6. An automatic inspection device for overhead power transmission lines, characterized in that: include: An acquisition module is used to obtain the point cloud data to be classified corresponding to the area to be inspected; There are ground objects and power lines in the area to be inspected; A classification module, used to classify the point cloud data to be classified using a lightweight power transmission line point cloud segmentation network, and obtain a first point cloud data set corresponding to the ground object and a second point cloud data set corresponding to the power line; A first processing module, used for processing the first point cloud data set based on a point cloud thinning algorithm to obtain a first comparison data set; A second processing module, used to determine the central axis point set data of the power line based on the second point cloud data set; A detection module, used to calculate the distance between each point in the first comparison data set and each point in the central axis point set data, and compare it with a preset safety distance to determine an inspection result; The acquisition submodule is used to obtain the point cloud data to be processed corresponding to the area to be inspected; A completion module, used to complete the point cloud data to be processed based on a point cloud completion algorithm, and obtain point cloud data to be classified; The first processing module is specifically used for: Divide a preset number of network blocks of the same spatial size based on the first point cloud data set; Determine the center point location corresponding to each network block; Determine, from among all the network blocks, a network block whose distance from the center point to the power line exceeds a first preset distance as a target network block; Randomly delete half of the point cloud data corresponding to the target network block, and integrate the remaining point cloud data to obtain a first comparison data set.

7. An automatic inspection device for overhead power transmission lines, characterized in that: include: Memory for storing computer programs; A processor, configured to implement the steps of the automatic inspection method for overhead power lines as claimed in any one of claims 1 to 5 when executing the computer program.

8. A readable storage medium, characterized in that: The readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the automatic inspection method of overhead transmission lines as claimed in any one of claims 1 to 5 are implemented.

Images