An intelligent planning method and system for the inspection route of power equipment

Abstract

This application relates to the field of path planning, and particularly to an intelligent inspection route planning method and system for power equipment. The method includes: dividing the power equipment environment into grids and planning an initial inspection path based on three-dimensional point cloud data. After the real-time environmental data is matched with the grids, obstacles are identified through connected component analysis, and the grid collision risk value is calculated. The trajectory time sequence length of the DWA algorithm is dynamically adjusted using local peak grids to predict multiple trajectories. The comprehensive risk degree is obtained by weighted averaging the collision risk values of the trajectory grid sequences, and the dynamic feedback coefficient is generated by mapping the risk difference between adjacent moments through a negative exponential function to correct the DWA evaluation function. Finally, the trajectory corresponding to the maximum dynamic avoidance evaluation function is selected to update the inspection route, realizing intelligent path planning. By introducing a dynamic feedback coefficient and collision risk assessment, this application improves the applicability and reliability of the algorithm in different scenarios.

Description

Technical Field

[0001] This application relates to the field of path planning, and specifically to an intelligent planning method and system for the inspection route of power equipment. Background Art

[0002] Power equipment is a key infrastructure for energy transmission and distribution in modern society, and is applied to various links such as power generation, power transmission, power transformation, and power distribution. The stable operation of these devices is crucial for ensuring the reliability and safety of the power system. Power equipment usually includes complex and high-value facilities such as high-voltage transmission lines, substations, transformers, switchgear, and cables, which are distributed in a vast geographical area and face dual challenges from the natural environment and human factors. With the continuous expansion of the scale of the power system and the increasing technological complexity, the management and maintenance of power equipment have become even more critical.

[0003] The inspection of power equipment is an important link to ensure its normal operation. Traditional inspection methods usually rely on manual labor, with low efficiency and being easily affected by human factors. With the development of technology, intelligent inspection technology has gradually emerged, and the intelligent planning of inspection routes is the core component of the intelligent inspection system. Currently, the operating environment of power equipment is complex and changeable, with many dynamic obstacles. These obstacles not only interfere with the normal operation of inspection equipment but may also cause collision accidents, affecting the continuity of inspection work and the service life of the equipment.

[0004] Currently, the DWA algorithm is usually used to plan local inspection routes for power equipment. However, the traditional DWA algorithm uses a fixed evaluation function to evaluate the advantages and disadvantages of different trajectories, resulting in the inability to timely reflect the changes of dynamic obstacles in the complex and changeable inspection environment of power equipment, which may lead to a collision risk in the planned path and affect the safety of inspection equipment; at the same time, it is unable to effectively utilize real-time environmental information for path optimization, which may lead to unreasonable path planning, increasing inspection time and resource consumption. Summary of the Invention

[0005] In order to solve the problem that the traditional DWA algorithm, due to using a fixed evaluation function, cannot adapt to the changes of dynamic obstacles in the inspection of power equipment, resulting in a collision risk in path planning and increasing inspection time and resource consumption, this application provides an intelligent planning method and system for the inspection route of power equipment.

[0006] According to the first aspect of the present application, an intelligent planning method for the inspection route of power equipment is provided, including: obtaining environmental data of the area where the power equipment is located according to three-dimensional point cloud data, dividing the environmental data into grid data, and using the grid data to plan the initial path of the power equipment inspection; obtaining real-time environmental data of the area to be inspected by the power equipment, matching the real-time environmental data with the grid data to obtain dynamic data and performing connected domain analysis, identifying obstacle data, and obtaining the collision risk value of the grid according to the time difference and position relationship between the obstacle data and the grid and the dynamic characteristics of the obstacle data; obtaining the local peak grid of the collision risk value, dynamically adjusting the timing length of each trajectory in the DWA algorithm using the local peak grid, obtaining the dynamic timing length of the adjusted multi-trajectory, and performing multi-trajectory prediction; obtaining the grid sequence and collision risk value passed by each trajectory, obtaining the comprehensive risk degree by taking the weighted average in the order of the position of the grid in the trajectory, mapping the difference in the comprehensive risk degree between adjacent moments with a negative exponential function, obtaining the dynamic feedback coefficient and correcting the evaluation function of the DWA algorithm to obtain a dynamic avoidance evaluation function; based on the dynamic avoidance evaluation function, selecting the trajectory corresponding to the value of the maximum dynamic avoidance evaluation function as the result of local path planning, and updating the inspection route of the inspection equipment according to the result of local path planning to complete the intelligent planning of the inspection route of the power equipment.

[0007] Preferably, the dividing the environmental data into grid data includes:

[0008] Constructing a grid system in three-dimensional space based on a preset side length, using the grid system as a dividing unit, and performing equally spaced grid division on the environmental data with the dividing unit to obtain a number of grid units in the area to be planned.

[0009] By constructing a grid system in three-dimensional space based on a preset side length and performing equally spaced division on the environmental data as a dividing unit, the complex environmental data structure is structured into a number of ordered grid units. This not only simplifies the data processing process and improves the calculation efficiency, but also clearly distinguishes the obstacle area from the passable area, providing an accurate basis for the initial path planning.

[0010] Preferably, the initial path includes:

[0011] In the divided grid data, marking the grids belonging to static obstacles as obstacle coordinates, and marking the corresponding grids of the power equipment to be inspected in the grid data as inspection coordinate points;

[0012] Determining the starting coordinate and ending coordinate of the inspection equipment, and using a path planning algorithm to obtain a feasible path from the starting point to the ending point, which is used as the initial path of the power equipment inspection.

[0013] Preferably, the obtaining the dynamic data includes:

[0014] Match the real-time environment data with the grid data using a point cloud registration algorithm to obtain the registered real-time environment data; perform an exclusive OR operation on the registered real-time point cloud data and the grid data to obtain dynamic data.

[0015] Preferably, the identifying obstacle data includes:

[0016] Perform connected component analysis on the dynamic data to obtain the connected component of the point cloud data corresponding to each dynamic object, remove the dynamic data with the volume of the connected component of the point cloud data less than the preset volume threshold, and the remaining connected components correspond to dynamic obstacles, and use the data of each connected component as the data of the corresponding obstacle.

[0017] By analyzing the connected components of the dynamic data, accurately identify the region of the point cloud data corresponding to each dynamic object, remove the connected components with too small volume to filter out noise and invalid information, and the remaining connected components correspond to real dynamic obstacles, which can accurately extract the dynamic obstacle data that needs attention from the complex environment.

[0018] Preferably, the obtaining the collision risk value of the grid includes:

[0019] Perform connected component analysis on the point cloud data of each obstacle to obtain the geometric center of a single obstacle, and use the historical coordinates of the geometric center at the previous preset number of time points of the obstacle as the historical movement trajectory of the obstacle, perform trajectory prediction on the historical movement trajectory of the obstacle to obtain the trajectory prediction result of the obstacle;

[0020] Obtain the volume data of multiple time points in the historical movement trajectory of the obstacle to form a volume sequence, perform a first-order difference on the volume sequence to obtain the volume change amount between adjacent time points, and calculate the mean value of the first-order difference result to obtain the average volume change amount of the obstacle; use an exponential function to map the product of the average volume change amount and the hyperparameter to obtain the volume risk coefficient;

[0021] Calculate the time difference between the time point when the inspection device reaches the th grid and the time point when the th obstacle reaches the th grid, and adjust the weight of the time difference by multiplying it by the hyperparameter; use an exponential function to map the absolute value or square of the adjusted time difference to obtain the time risk coefficient.

[0022] Respectively obtain the time points when the inspection device and the obstacle reach the th grid, compare who reaches the th grid first between the inspection device and the obstacle to determine the first-arrival object and the late-arrival object;

[0023] In response to the inspection device reaching the If the time point of a grid is less than or equal to the time point when the obstacle reaches the th grid, the inspection device is the first to reach the object; otherwise, the obstacle is the first to reach the object.

[0024] Calculate the position coordinate point of the later-arriving object when the first-arriving object reaches the th grid, calculate the Euclidean distance between the position coordinate point of the later-arriving object and the center point of the th grid, and use the exponential function to map the product of the Euclidean distance and the hyperparameter to obtain the distance risk coefficient.

[0025] Preferably, the peak grid for obtaining the collision risk value includes:

[0026] Use a voxel window to slide over the point cloud data to cover the entire area to be inspected. Among them, the voxel window is a cube with a preset number of grids, and the sliding step is usually the same as the side length of the voxel.

[0027] If the collision risk value of the central grid of the voxel window is greater than or equal to the collision risk values of all grids within the neighborhood of the central grid, then the central grid of the voxel window is a local peak point grid; otherwise, it is not a local peak point grid.

[0028] Preferably, the dynamic time series length includes:

[0029] Calculate the Euclidean distance between the three-dimensional coordinate point of the current inspection device and the central grid of each local peak point grid, and use the local peak point grid with the smallest Euclidean distance as the risk grid.

[0030] Calculate the Euclidean distance between the three-dimensional coordinate point of the current inspection device and the three-dimensional coordinate point of the risk grid, use the negative exponential function to map the Euclidean distance to obtain the spatial distance between the current inspection device and the risk grid, and use the sum of the spatial distance and the collision risk value as the risk index.

[0031] Calculate the cosine similarity between the relative angle value between the current inspection device and the risk grid and the relative angle values between each trajectory and the risk grid.

[0032] Use the product of the initial time series length, the risk index, the cosine similarity, and the maximum value of the preset adjusted time series length as the time series prediction length value of each trajectory.

[0033] In the second aspect of the present application, there is also provided an intelligent planning system for the inspection route of power equipment, including a processor and a memory. The memory stores computer program instructions, and when the computer program instructions are executed by the processor, an intelligent planning method for the inspection route of power equipment according to the first aspect of the present application is implemented.

[0034] The technical solution of this application has the following beneficial technical effects: By obtaining environmental data in real time and dynamically adjusting the evaluation function, it can timely reflect the changes of dynamic obstacles in the environment, effectively avoid obstacles, reduce the collision risk, and ensure the safe operation of the inspection equipment; The introduction of dynamic feedback coefficients and collision risk assessment enables the algorithm to adapt to the complex and changeable power equipment inspection environment, flexibly adjust the path planning strategy, and improve the applicability and reliability of the algorithm in different scenarios; Utilizing real-time environmental information and dynamic adjustment mechanisms, a better inspection path can be generated, reducing unnecessary path adjustments and repeated inspections, thereby saving inspection time and resources and improving inspection efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 is a flowchart of an intelligent inspection route planning method for a power equipment according to an embodiment of this application.

[0036] Figure 2 is a structural block diagram of an intelligent inspection route planning system for a power equipment according to an embodiment of this application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0037] Next, the technical solutions in the embodiments of this application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are some, but not all, of the embodiments of this application. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative efforts shall fall within the protection scope of this application.

[0038] According to the first aspect of this application, this application provides an intelligent inspection route planning method and system for a power equipment. Figure 1 is a flowchart of an intelligent inspection route planning method for a power equipment according to an embodiment of this application. As Figure 1 shown, the intelligent inspection route planning method for the power equipment includes steps S1 to S5, which are described in detail below.

[0039] Step S1: Obtain environmental data of the area where the power equipment is located according to the three-dimensional point cloud data, divide the environmental data into grid data, and use the grid data to plan the initial path of the power equipment inspection.

[0040] It should be noted that a three-dimensional point cloud data acquisition device (such as a laser scanner, etc.) is used to scan the area where the power equipment's path needs to be planned to obtain the three-dimensional point cloud data in this area. These data contain the spatial position information of each point in the area.

[0041] Based on a preset side length, a grid system is constructed in the three-dimensional space. Using the grid system as the division unit, the environmental data is equally spaced grid-divided using the division unit to obtain several grid units in the area to be planned.

[0042] Exemplarily, a cube with a preset side length of 0.05 meters can be adjusted according to specific implementation situations.

[0043] In the divided grid data, the grids belonging to static obstacles are marked as obstacle coordinates, and the corresponding grids of the power equipment to be inspected in the grid data are marked as inspection coordinate points; the specific positions of obstacles and inspection targets in the environment are clarified, providing necessary information for path planning.

[0044] Determine the starting coordinate and ending coordinate of the inspection device, use a path planning algorithm to obtain a feasible path from the starting point to the ending point, and use it as the initial path for power equipment inspection.

[0045] Existing path planning methods include, but are not limited to: path planning methods such as the A-star algorithm and the Dijkstra algorithm, which can effectively find the optimal path in a grid environment. This technology is well-known to those skilled in the art and will not be described in detail.

[0046] It should be noted that although the initial path planning provides the basic route for inspection, during the actual inspection process, the environment may change and dynamic factors (such as moving obstacles, etc.) may appear. To cope with these dynamic factors, local path planning will be carried out in the subsequent steps to avoid dynamic factors and ensure that the inspection device can complete the inspection task safely and efficiently.

[0047] Step S2: Obtain the real-time environmental data of the area to be inspected of the power equipment, match the real-time environmental data with the grid data to obtain dynamic data and perform connected component analysis, identify obstacle data, and obtain the collision risk value of the grid according to the time difference and position relationship between the obstacle data and the grid and the dynamic characteristics of the obstacle data.

[0048] It should be noted that a depth camera is installed in the area to be inspected of the power equipment to scan the surrounding environment at a frequency of 10 times per second to obtain three-dimensional point cloud data with depth information; at the same time, the position and attitude information of the inspection device is recorded through an inertial navigation unit (IMU) installed on the inspection device.

[0049] Obtain dynamic data, including:

[0050] Use a point cloud registration algorithm to match the real-time environmental data with the grid data to obtain the registered real-time environmental data; perform an exclusive OR operation on the registered real-time point cloud data and the grid data to obtain dynamic data.

[0051] It should be noted that the point cloud registration algorithms include: ICP (Iterative Closest Point) algorithm, GMM (Gaussian Mixture Model) algorithm, and CPD (Coherent Point Drift) algorithm, which are all well-known technologies in the field and will not be described in detail.

[0052] Among them, the rules of the exclusive OR operation are as follows: if the two input values are the same, the output is 0; if the two input values are different, the output is 1.

[0053] In the processing of point cloud data, the existence of each point can be regarded as a logical value (0 or 1). Through the exclusive OR operation, the different parts between two point cloud data sets can be identified.

[0054] Exemplarily, the point cloud : represents grid data (static environment); the point cloud : represents real-time point cloud data (dynamic environment). Each point cloud data is represented as a set, where the position coordinates of each point are an element in the set. For example: the point cloud : ; the point cloud : , and the point cloud data is divided into grids. Each grid cell can be represented as a binary value: if there is point cloud data in the grid cell, the value is 1; if there is no point cloud data in the grid cell, the value is 0.

[0055] Perform the exclusive OR operation on each grid. If the grid cell is 1 in the point cloud and 0 in the point cloud, or vice versa, the exclusive OR operation result of this grid cell is 1; if the grid cell is 1 or 0 in both the point cloud and the point cloud, the exclusive OR operation result of this grid cell is 0. Through the exclusive OR operation, a point cloud data set with table differences can be obtained, where the point cloud data set contains the parts that have changed between the point cloud and the point cloud, which are the dynamic data.

[0056] Exemplarily, the point cloud (static environment): ; the point cloud (dynamic environment): ; divide these two point cloud data into grids and perform the exclusive OR operation: grid : has, has, then it is 0; grid : has, has, then it is 0; grid : has, has none, then it is 1; grid : None in If there is in, it is 1; the XOR result dynamic data is: .

[0057] In the inspection of power equipment, dynamic obstacles (such as moving personnel, vehicles, etc.) can be quickly identified through XOR operations, thus providing data support for subsequent collision risk assessment and path planning.

[0058] Identifying obstacle data includes:

[0059] Perform connected component analysis on the dynamic data to obtain the connected component of the point cloud data corresponding to each dynamic object, remove the dynamic data with the volume of the connected component of the point cloud data less than the preset volume threshold, and the remaining connected components correspond to dynamic obstacles. Take the data of each connected component as the data of the corresponding obstacle.

[0060] Exemplarily, the preset volume threshold is 0.5 cubic meters, and the implementer can adjust it according to the specific situation.

[0061] Obtaining the collision risk value of the grid includes:

[0062] Perform connected component analysis on the point cloud data of each obstacle to obtain the geometric center of a single obstacle, and use the historical coordinates of the geometric center at the previous preset number of time points of the obstacle as the historical running trajectory of the obstacle. Perform trajectory prediction on the historical running trajectory of the obstacle to obtain the trajectory prediction result of the obstacle.

[0063] Obtain the volume data of multiple time points in the historical running trajectory of the obstacle to form a volume sequence, perform a first-order difference on the volume sequence to obtain the volume change amount between adjacent time points, and calculate the mean value of the first-order difference result to obtain the average volume change amount of the obstacle.

[0064] Use an exponential function to map the product of the average volume change amount and the hyperparameter to obtain the volume risk coefficient.

[0065] Specifically, the volume risk coefficient satisfies the following relational expression:

[0066] ;

[0067] In the formula, represents the volume risk coefficient of the th obstacle in the th grid, represents the hyperparameter, represents the first-order difference result of the volume sequence of the th obstacle, represents the mean value function, represents the exponential function with the natural number as the base.

[0068] That is to say, the hyperparameter , can be adjusted according to the specific scenario, mapping the volume change amount to a positively correlated risk value. The larger the volume risk coefficient, the higher the collision risk of the obstacle.

[0069] Calculate the time difference between the time point when the inspection device reaches the th grid and the time point when the th obstacle reaches the th grid, and multiply the time difference by the hyperparameter for weight adjustment;

[0070] Use the exponential function to map the absolute value or square of the adjusted time difference to obtain the time risk coefficient.

[0071] Specifically, the time risk coefficient satisfies the following relational expression:

[0072] ; or

[0073] ;

[0074] In the formula, represents the time risk coefficient of the th obstacle in the th grid, represents the hyperparameter, represents the time point when the inspection device reaches the th grid, represents the th obstacle reaches the th grid.

[0075] Exemplarily, the hyperparameter , can be adjusted according to the specific scenario. The smaller the time difference, the higher the risk.

[0076] Obtain the time points when the inspection device and the obstacle reach the th grid respectively, compare which one of the inspection device and the obstacle reaches the th grid first, and determine the first-arrival object and the late-arrival object;

[0077] In response to the time point when the inspection device reaches the th grid being less than or equal to the time point when the obstacle reaches the th grid, the inspection device is the first-arrival object, otherwise the obstacle is the first-arrival object;

[0078] Calculate the position coordinate point of the late-arrival object when the first-arrival object reaches the th grid, and calculate the position coordinate point of the late-arrival object and the The Euclidean distance from the center point of a grid is used to map the product of the Euclidean distance and a hyperparameter using an exponential function to obtain a distance risk coefficient.

[0079] Specifically, the distance risk coefficient satisfies the following relational expression:

[0080] ;

[0081] In the formula, represents the distance risk coefficient of the th obstacle in the th grid, represents the hyperparameter, represents the coordinate point of the late-arriving object, represents the coordinate point of the center point of the th grid, represents the exponential function with the natural number as the base.

[0082] Exemplarily, the hyperparameter , which can be specifically adjusted according to the implementation situation. The smaller the distance, the higher the risk.

[0083] Specifically, the collision risk value satisfies the following relational expression:

[0084] ;

[0085] In the formula, represents the collision risk value of the th obstacle in the th grid, represents the volume risk coefficient of the th obstacle, represents the time risk coefficient of the th obstacle in the th grid, represents the distance risk coefficient of the th obstacle in the th grid.

[0086] Perform a point cloud dilation operation on the grid occupied by the initial path, and use the grids with the preset number of dilated grids as selectable grids; calculate the Euclidean distance between the geometric center of the obstacle and the real-time position of the inspection device, and in response to the Euclidean distance being less than or equal to the preset distance, regard the obstacle as an obstacle that may cause a collision.

[0087] Exemplarily, the number of dilated grids is 30, which can be adjusted according to the specific scenario; the preset distance is 10 meters, which can be adjusted according to the specific scenario.

[0088] Step S3: Obtain the local peak grid of the collision risk value, and use the local peak grid to dynamically adjust the temporal lengths of each trajectory in the DWA algorithm to obtain the dynamic temporal lengths of the adjusted multi-trajectories, and perform prediction on the multi-trajectories.

[0089] By analyzing the risk factors in the environment, dynamically adjust the temporal length in the DWA algorithm to ensure that the inspection equipment can complete the task safely and efficiently.

[0090] First, obtain the peak grid of the collision risk value. The local risk value grid represents the area with the highest risk in the environment, specifically including:

[0091] Use a voxel window to slide over the point cloud data to cover the entire area to be inspected. Among them, the voxel window is a cube with a preset number of grids, and the sliding step is usually the same as the side length of the voxel.

[0092] Exemplarily, the voxel window is a cube of a 26-neighborhood grid, and one voxel corresponds to one cube in Step S1.

[0093] In response to the collision risk value of the central grid of the voxel window being greater than or equal to the collision risk values of all grids within the neighborhood of the central grid, the central grid of the voxel window is the local peak point grid; otherwise, it is not the local peak point grid.

[0094] That is to say, during the sliding process, compare the collision risk value of the central grid of the voxel window with the collision risk values of all grids within its neighborhood. If the value of the central grid is the largest, it is marked as the local peak point grid.

[0095] The dynamic temporal length includes:

[0096] Calculate the Euclidean distance between the three-dimensional coordinate point of the current inspection equipment and the central grid of each local peak point grid, and take the local peak point grid with the smallest Euclidean distance as the risk grid.

[0097] Calculate the Euclidean distance between the three-dimensional coordinate point of the current inspection equipment and the three-dimensional coordinate point of the risk grid, use the negative exponential function to map the Euclidean distance to obtain the spatial distance between the current inspection equipment and the risk grid, and take the sum of the spatial distance and the collision risk value as the risk index.

[0098] Calculate the cosine similarity between the relative angle value between the current inspection equipment and the risk grid and the relative angle values between each trajectory and the risk grid.

[0099] Take the product of the initial temporal length, the risk index, the cosine similarity, and the maximum value of the preset adjusted temporal length as the dynamic temporal length of each trajectory.

[0100] Specifically, the dynamic temporal length satisfies the following relational expression:

[0101] ;

[0102] In the formula, represents the dynamic time series length of the th trajectory, represents the initial time series length, represents the Euclidean distance value between the three-dimensional coordinate point of the current inspection device and the three-dimensional coordinate point of the risk grid, represents the collision risk value of the risk grid, represents the maximum value of the preset adjustment time series length, represents the cosine similarity between the relative angle value between the current inspection device and the risk grid during real-time inspection and the relative angle value between the th trajectory and the risk grid, represents the exponential function with the natural number as the base.

[0103] That is to say, it is a comprehensive risk assessment index used to measure the overall risk of the inspection device when approaching a certain risk grid. The value of decreases as the Euclidean distance increases, reflecting that the farther the distance, the lower the risk; is the collision risk value of the risk grid, directly reflecting the possibility of collision of the grid.

[0104] Step S4: Obtain the grid sequence and collision risk value passed by each trajectory, and calculate the weighted average according to the position order of the grids in the trajectory to obtain the risk comprehensive degree. Map the difference in the risk comprehensive degree between adjacent moments with the negative exponential function to obtain the dynamic feedback coefficient and correct the evaluation function of the DWA algorithm to obtain the dynamic avoidance evaluation function.

[0105] Specifically, the dynamic feedback coefficient satisfies the following relational formula:

[0106] ;

[0107] ;

[0108] In the formula, represents the risk comprehensive degree of the th trajectory, represents the total number of grids passed by the trajectory, represents the weight of the th grid, represents the collision risk value of the th grid, , respectively represent the risk comprehensive degrees of the trajectory at two adjacent moments, represents the exponential function with the natural number as the base, Indicates The dynamic feedback coefficient of the trajectory.

[0109] Specifically, the dynamic avoidance evaluation function satisfies the following relationship:

[0110] ;

[0111] In the formula, Indicates The dynamic avoidance evaluation function of the trajectory is: Indicates The dynamic feedback coefficient of the trajectory, Indicates The initial evaluation function of the trajectory.

[0112] It should be noted that the weights of grids close to the starting point of the trajectory are higher because they are closer to the current moment and have a greater impact on the immediate risk. The collision risk value of each grid is multiplied by its weight and summed to obtain the comprehensive risk level of the trajectory.

[0113] The characteristic of the negative exponential function is that when the difference is small, the mapping value is large; when the difference is large, the mapping value is small. The dynamic feedback coefficient obtained in this way can reflect the rate of risk change. The smaller the difference, the slower the risk change and the larger the feedback coefficient; the larger the difference, the more drastic the risk change and the smaller the feedback coefficient.

[0114] The original evaluation function is usually based on factors such as path length and smoothness, while the dynamic feedback coefficient introduces dynamic adjustment of risk assessment. In this way, the dynamic avoidance evaluation function not only considers the geometric characteristics of the trajectory, but also integrates the risk factors in the dynamic environment, making path planning more intelligent and safer.

[0115] The dynamic feedback coefficient adjusts the original evaluation function according to the risk change, making the evaluation result more consistent with the actual situation in the dynamic environment. It effectively combines risk assessment with path planning, improving the adaptability and safety of inspection equipment in complex environments.

[0116] Step S5: Based on the dynamic avoidance evaluation function, select the trajectory corresponding to the maximum dynamic avoidance evaluation function value as the result of local path planning, update the inspection route of the inspection equipment according to the result of local path planning, and complete the intelligent planning of the power equipment inspection route.

[0117] Through the above process, the intelligent planning of the entire power equipment inspection route is completed. It not only takes into account obstacles in the static environment, but also dynamically adapts to changes in the environment, ensuring that the inspection equipment can complete its tasks safely and efficiently in a complex and changing environment.

[0118] Thus, an intelligent planning method for the inspection route of a power equipment is completed.

[0119] The embodiment of the present application also discloses an intelligent planning system for the inspection route of a power equipment. As Figure 2 shown, the system includes a processor and a memory. The memory stores computer program instructions, and when the computer program instructions are executed by the processor, an inspection method for part processing according to the first aspect of the present invention is implemented. The system also includes a communication bus, a communication interface and other components well known to those skilled in the art. Their settings and functions are known in the art, so they will not be described in detail here. The above system also includes a communication bus, a communication interface and other components well known to those skilled in the art. Their settings and functions are known in the art, so they will not be described in detail here.

[0120] Although this specification has shown and described multiple embodiments of the present application, it is obvious to those skilled in the art that such embodiments are provided only by way of example. Those skilled in the art will think of many changes, alterations and alternative ways without departing from the spirit and scope of the present application. It should be understood that various alternative solutions of the embodiments of the present application described herein can be adopted in the process of practicing the present application.

[0121] The above are all the preferred embodiments of the present application. The protection scope of the present application is not limited accordingly. Therefore, all equivalent changes made according to the structure, shape and principle of the present application should be covered within the protection scope of the present application.

Claims

1. A method for intelligent planning of inspection routes for power equipment, characterized in that: include: Obtain environmental data of the area where the power equipment is located based on the three-dimensional point cloud data, divide the environmental data into grid data, and use the grid data to plan the initial path for power equipment inspection; Obtain real-time environmental data of the area to be inspected for power equipment, match the real-time environmental data with the grid data to obtain dynamic data and perform connected domain analysis, identify obstacle data, and obtain the collision risk value of the grid based on the time difference and position relationship between the obstacle data and the grid and the dynamic characteristics of the obstacle data; Obtain the local peak grid of the collision risk value, use the local peak grid to dynamically adjust the time series length of each track in the DWA algorithm, obtain the dynamic time series length of the adjusted multiple tracks, and predict the multiple tracks; Obtain the grid sequence and collision risk value of each trajectory, take the weighted average according to the position order of the grids in the trajectory to get the comprehensive risk degree, use the negative exponential function to map the difference of the comprehensive risk degree at adjacent moments, get the dynamic feedback coefficient and correct the evaluation function of the DWA algorithm to obtain the dynamic avoidance evaluation function; Based on the dynamic avoidance evaluation function, the trajectory corresponding to the maximum dynamic avoidance evaluation function value is selected as the result of local path planning, and the inspection route of the inspection equipment is updated according to the result of local path planning to complete the intelligent planning of the inspection route of the power equipment; Specifically, obtaining the collision risk value of the grid includes: Perform connected domain analysis on the point cloud data of each obstacle to obtain the geometric center of a single obstacle, and use the historical coordinates of the geometric center at a preset number of moments before the obstacle as the historical running trajectory of the obstacle, perform trajectory prediction on the historical running trajectory of the obstacle, and obtain the trajectory prediction result of the obstacle; The volume data of multiple time points in the historical running trajectory of the obstacle are obtained to form a volume sequence. The volume sequence is first-order differentiated to obtain the volume change at adjacent time points, and the mean of the first-order difference results is calculated to obtain the average volume change of the obstacle. The exponential function is used to map the product of the average volume change and the hyperparameter to obtain the volume risk coefficient. Calculate the inspection equipment to arrive at the The time point of the grid and the Obstacles reach the The time difference between the time points of the grids is multiplied by the hyperparameter to adjust the weight; the absolute value or square of the adjusted time difference is mapped using an exponential function to obtain the time risk coefficient; Get the inspection equipment and obstacles to reach the first At the time point of each grid, compare which one arrives first, the inspection equipment or the obstacle. grids to determine the objects that arrive first and the objects that arrive later; In response to the inspection equipment arriving at the The time point of the grid is less than or equal to the time when the obstacle reaches the If the time point of a grid is less than 1, the inspection equipment will reach the object first, otherwise the obstacle will reach the object first. Calculate the first arriving object to arrive at When there are grids, the position coordinates of the late arriving object are calculated. The Euclidean distance of the center points of the grids is mapped using an exponential function to the product of the Euclidean distance and the hyperparameter to obtain the distance risk coefficient; The total of the volume risk coefficient, time risk coefficient and distance risk coefficient multiplied together is taken as the collision risk value.

2. A method for intelligently planning inspection routes for electric power equipment according to claim 1, characterized in that: The step of dividing the environmental data into grid data includes: A grid system is constructed in three-dimensional space based on a preset side length. The grid system is used as a division unit. The environmental data is divided into equally spaced grids using the division units to obtain a number of grid units in the area to be planned.

3. A method for intelligently planning inspection routes for electric power equipment according to claim 1, characterized in that: The initial path includes: In the divided grid data, the grids belonging to static obstacles are marked as obstacle coordinates, and the corresponding grids of the power equipment to be inspected in the grid data are marked as inspection coordinate points; Determine the starting point and end point coordinates of the inspection equipment, use the path planning algorithm to obtain a feasible path from the starting point to the end point, and use it as the initial path for power equipment inspection.

4. A method for intelligently planning inspection routes for electric power equipment according to claim 1, characterized in that: The obtaining of dynamic data comprises: The real-time environment data is matched with the grid data using a point cloud registration algorithm to obtain the registered real-time environment data; the registered real-time point cloud data is XOR-ed with the grid data to obtain dynamic data.

5. A method for intelligent planning of inspection routes for electric power equipment according to claim 1, characterized in that: The obstacle identification data includes: The connected domain analysis is performed on the dynamic data to obtain the connected domain of the point cloud data corresponding to each dynamic object. The dynamic data whose connected domain volume of the point cloud data is less than the preset volume threshold is eliminated. The remaining connected domains correspond to dynamic obstacles, and the data of each connected domain is used as the data of the corresponding obstacle.

6. A method for intelligently planning inspection routes for electric power equipment according to claim 1, characterized in that: The step of obtaining a peak grid of collision risk values ​​includes: Use a voxel window to slide the point cloud data to cover the entire area to be inspected, where the voxel window is a cube with a preset number of grids, and the sliding step length is usually consistent with the side length of the voxel; In response to the collision risk value of the central grid of the voxel window being greater than or equal to the collision risk values ​​of all grids in the neighborhood of the central grid, the central grid of the voxel window is a local peak point grid, otherwise, it is not a local peak point grid.

7. A method for intelligently planning inspection routes for electric power equipment according to claim 1, characterized in that: The dynamic time sequence length includes: Calculate the Euclidean distance between the three-dimensional coordinate point of the current inspection equipment and the central grid of each local peak point grid, and take the local peak point grid with the smallest Euclidean distance as the risk grid; Calculate the Euclidean distance between the three-dimensional coordinate point of the current inspection device and the three-dimensional coordinate point of the risk grid, use the negative exponential function to map the Euclidean distance, obtain the spatial distance between the current inspection device and the risk grid, and use the sum of the spatial distance and the collision risk value as the risk indicator; Calculate the cosine similarity of the relative angle value between the current inspection device and the risk grid and the relative angle value between each trajectory and the risk grid; The product of the initial time series length, the risk index and the cosine similarity and the maximum value of the preset adjusted time series length is taken as the time series prediction length value of each trajectory.

8. An intelligent inspection route planning system for power equipment, characterized in that: include: processor; and a memory storing computer instructions for a flaw detection method for bowl forgings based on image recognition, wherein when the computer instructions are executed by the processor, the device executes an intelligent inspection route planning method for electric power equipment according to any one of claims 1-7.

Images