Inspection path planning method and system of substation inspection unmanned aerial vehicle

Abstract

The invention discloses a routing inspection path planning method and system for a substation routing inspection unmanned aerial vehicle, and the method comprises the steps: obtaining the point cloud data of a substation, and constructing a three-dimensional point cloud model of the substation; receiving an inspection instruction for the target inspection main body, obtaining a starting point and an ending point of the unmanned aerial vehicle, and generating an initial inspection path; power frequency electric field intensity data and power frequency magnetic field intensity data of a to-be-measured point of the transformer substation are received, the transformer substation is divided into m electromagnetic field areas, and the m electromagnetic field areas comprise at least one strong electromagnetic field area; communication parameters of the unmanned aerial vehicle are obtained, and the safe operation distance of the unmanned aerial vehicle in each strong electromagnetic field area is determined; according to the three-dimensional point cloud model and the safe operation distance of each strong electromagnetic field area, adjusting the initial inspection path to obtain a final inspection path; and sending the final inspection path to the unmanned aerial vehicle. The method is used for effectively reducing the out-of-control risk of the unmanned aerial vehicle in a substation in a complex electromagnetic environment, and the safety is improved.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of path planning, and in particular to a method and system for planning an inspection path for a substation inspection drone. Background Art

[0002] With the continuous development and expansion of power systems, substation inspection work has become increasingly important and complex. Traditional manual inspection methods face problems such as low efficiency, high safety risks, and limited coverage. In order to solve these problems, drone inspection technology has emerged and has become an important means of substation inspection. Drone inspection has the advantages of high flexibility, high efficiency, and the ability to reach areas that are difficult for humans to reach, which can greatly improve the quality and efficiency of inspections.

[0003] However, the substation environment is complex, with a large number of high-voltage equipment and complex electromagnetic environment, which brings huge challenges to drone inspection. The strong electromagnetic field in the substation will not only affect the communication and navigation systems of the drone, but may also cause damage to the drone itself. Therefore, how to ensure the safe operation of the drone while ensuring the inspection effect has become an important issue facing substation drone inspection technology.

[0004] Traditional UAV route planning methods focus on obstacle avoidance and coverage. These methods are usually based on a 3D model or map of the substation and use various algorithms to generate inspection paths. Common algorithms include the artificial potential field method, which sets the target point as an attractive potential field and the obstacle as a repulsive potential field. The UAV moves in this virtual potential field, and is affected by the combined effects of attractive and repulsive forces, eventually forming a path that avoids obstacles.

[0005] However, traditional drone route planning methods have obvious shortcomings when dealing with substations with complex electromagnetic environments. They do not take into account the complex electromagnetic environment in substations. Strong electromagnetic field areas may interfere with the drone's communication system, resulting in loss of control signals or reduced communication quality, increasing the risk of the drone losing control.

[0006] Therefore, new path planning methods need to be developed to achieve safer and more efficient substation drone inspections. Summary of the invention

[0007] The embodiments of the present application provide a method and system for planning an inspection path for a substation inspection drone, which are used to effectively reduce the risk of the drone losing control in a substation with a complex electromagnetic environment and improve safety.

[0008] To achieve the above objectives, the embodiments of the present application adopt the following technical solutions:

[0009] In a first aspect, a method for planning an inspection path of a substation inspection drone is provided, which is applied to a central control host, the central control host is connected to a drone, and an electromagnetic detection device is provided on the drone, the electromagnetic detection device is used to detect the power frequency electric field strength and the power frequency magnetic field strength, and the method comprises:

[0010] Acquiring point cloud data of the substation through the drone, and constructing a three-dimensional point cloud model of the substation based on the point cloud data;

[0011] In response to receiving an inspection instruction for a target inspection subject, obtaining a starting point and an end point of the drone, and generating an initial inspection path according to the three-dimensional point cloud model, the starting point, the end point and a preset flight altitude, wherein the target inspection subject is located on the initial inspection path;

[0012] In response to receiving the power frequency electric field strength data and the power frequency magnetic field strength data of the test point of the substation transmitted by the drone, the substation is divided into m electromagnetic field areas according to the power frequency electric field strength data and the power frequency magnetic field strength data of the test point, where m is an integer greater than 1, and the m electromagnetic field areas include at least one strong electromagnetic field area, at least one standard electromagnetic field area and / or at least one weak electromagnetic field area;

[0013] Acquire the communication parameters of the drone, and determine the safe operating distance of the drone in each of the strong electromagnetic field areas according to the power frequency electric field strength data, the power frequency magnetic field strength data and the communication parameters of each of the strong electromagnetic field areas;

[0014] According to the three-dimensional point cloud model and the safe operating distance of each of the strong electromagnetic field areas, the initial inspection path is adjusted to obtain a final inspection path;

[0015] The final inspection path is sent to the drone, so that the drone inspects the substation according to the final inspection path.

[0016] In a possible implementation manner of the first aspect, generating an initial inspection path according to the three-dimensional point cloud model, the starting point, the end point, and a preset flight altitude includes:

[0017] Performing semantic segmentation on the three-dimensional point cloud model using a point cloud processing network to segment the three-dimensional point cloud model into n semantic regions, wherein each of the semantic regions represents an element type;

[0018] Using a pattern matching algorithm to compare the n semantic regions with a pre-built standard template library of substation equipment structures, and determining the element features corresponding to each semantic region;

[0019] Determine the spatial relationship and functional relationship between each of the elements in the substation according to a preset knowledge graph;

[0020] Acquire a 2D design document of the substation, determine a layout rule of the substation according to the 2D design document, and determine a property of each of the elements according to the layout rule of the substation;

[0021] Mapping the element type and element features corresponding to each semantic region, the spatial relationship and functional relationship between each element, and the property of each element into the three-dimensional point cloud model to annotate each element in the three-dimensional point cloud model;

[0022] Mapping the target inspection subject with the elements marked in the three-dimensional point cloud model to determine the target element in the three-dimensional point cloud model;

[0023] Mapping the starting point, the end point and the flight altitude to the three-dimensional point cloud model to obtain a drone starting point and a drone end point in the three-dimensional point cloud model;

[0024] A preset path planning algorithm is used to generate an initial inspection path in the three-dimensional point cloud model according to the drone starting point, the drone end point and the target element.

[0025] In another possible implementation of the first aspect, the step of constructing the substation equipment structure standard template library includes:

[0026] Acquire a site image of the substation, perform feature extraction on the site image, and obtain geometric features and texture features of each element in the substation, wherein the elements include buildings and equipment;

[0027] Using a preset edge detection algorithm and a preset corner point detection algorithm to identify the outline and key points of each of the elements;

[0028] A standard template library for substation equipment structures is constructed based on the geometric features, texture features, contours and key points of each element.

[0029] In another possible implementation manner of the first aspect, dividing the substation into m electromagnetic field areas according to the power frequency electric field strength data and the power frequency magnetic field strength data of the point to be measured includes:

[0030] Using a preset interpolation algorithm, interpolate the substation according to the power frequency electric field strength data and the power frequency magnetic field strength data of all the points to be measured, to obtain an electromagnetic field strength distribution map of the substation, wherein the electromagnetic field strength distribution map includes a plurality of interpolation points and / or points to be measured;

[0031] The interpolation points and / or points to be measured of the power frequency electric field strength data greater than a preset electric field strength threshold and the power frequency magnetic field strength data greater than a preset magnetic field strength threshold in the electromagnetic field strength distribution diagram are summarized to obtain m electromagnetic field regions.

[0032] In another possible implementation manner of the first aspect, the communication parameters include communication signal frequency, antenna gain, bit error rate, and minimum acceptable signal strength, and determining the safe operating distance of the drone in each of the strong electromagnetic field areas according to the power frequency electric field strength data, the power frequency magnetic field strength data, and the communication parameters of each of the strong electromagnetic field areas includes:

[0033] For any of the strong electromagnetic field areas, the power frequency electric field strength data, the power frequency magnetic field strength data, the communication signal frequency and the antenna gain of each test point in the strong electromagnetic field area are input into a pre-constructed communication interference model to obtain the signal attenuation value of each test point of the drone in the strong electromagnetic field area;

[0034] Determine the interference threshold of the UAV in the strong electromagnetic field area according to the maximum signal attenuation value, the bit error rate and the minimum acceptable signal strength;

[0035] The safe operating distance of the UAV in the strong electromagnetic field area is determined according to the interference threshold.

[0036] In another possible implementation manner of the first aspect, determining the interference threshold of the drone in the strong electromagnetic field area according to the maximum signal attenuation value, the bit error rate, and the minimum acceptable signal strength includes:

[0037] Adding the minimum receivable signal strength to the maximum signal attenuation value to obtain the minimum acceptable strength of the UAV under the maximum attenuation condition;

[0038] Adding the minimum acceptable intensity, a preset impact factor, and a preset safety margin to obtain a critical value of interference of the UAV in the strong electromagnetic field area, wherein the impact factor is calculated based on the bit error rate;

[0039] Determining the safe operating distance of the UAV in the strong electromagnetic field area according to the interference threshold value includes:

[0040] Using a preset spatial path loss calculation formula, according to the interference critical value, the theoretical maximum communication distance of the UAV in the strong electromagnetic field area is calculated;

[0041] Determining the physical layout of the substation according to the three-dimensional point cloud model, and determining the environmental constraint distance of the drone according to the physical layout;

[0042] The minimum value of the theoretical maximum communication distance and the environmental constraint distance is multiplied by a preset safety factor to obtain a safe operating distance of the UAV in the strong electromagnetic field area.

[0043] In another possible implementation of the first aspect, adjusting the initial inspection path according to the three-dimensional point cloud model and the safe operating distance of each of the strong electromagnetic field areas to obtain a final inspection path includes:

[0044] Dividing the initial inspection path into a plurality of path segments according to the m electromagnetic field regions, and determining a target path segment in the strong electromagnetic field region among the plurality of path segments;

[0045] For each of the target path segments, construct a spherical safety area with the safe working distance as a radius in the three-dimensional point cloud model, wherein the center point of the target path segment is located at the center of the spherical safety area;

[0046] Determine whether the initial inspection path intersects with p spherical safety areas, where p is a positive integer;

[0047] If there is an intersection between the initial inspection path and p spherical safety areas, then determining a trajectory point on the initial inspection path that is within the spherical safety area;

[0048] For any of the track points, a corresponding replacement point is determined, and all of the track points are converted into the corresponding replacement points to obtain a final inspection path.

[0049] In another possible implementation manner of the first aspect, determining, for any one of the trajectory points, a replacement point corresponding thereto includes:

[0050] For any of the trajectory points, multiple candidate points are uniformly sampled on the spherical surface of the spherical safety area;

[0051] Among the plurality of candidate points, a point which is closest to the trajectory point and is not within any of the spherical safety regions is selected as a replacement point corresponding to the trajectory point.

[0052] In a second aspect, the present application provides a central control host, including:

[0053] a memory configured to store instructions; and

[0054] The processor is configured to call the instructions from the memory and implement the above-mentioned inspection path planning method for the substation inspection drone when executing the instructions.

[0055] In a third aspect, the present application provides a patrol path planning system for a substation patrol drone, comprising:

[0056] Central control host;

[0057] The drone is connected to the central control host, and the drone is provided with an electromagnetic detection device for detecting the intensity of the power frequency electric field and the power frequency magnetic field.

[0058] Through the above technical solution, the point cloud data of the substation is first obtained through the drone and a three-dimensional point cloud model is constructed, which provides an accurate spatial information basis for subsequent path planning. After receiving the inspection instruction, the initial inspection path is generated according to the starting point, end point, three-dimensional point cloud model and preset flight altitude of the drone to ensure that the target inspection subject is located on the path, ensuring the basic coverage rate of the inspection. Next, the electromagnetic detection device on the drone is used to collect the power frequency electric field strength and power frequency magnetic field strength data in real time, and the substation is divided into multiple electromagnetic field areas, including strong electromagnetic field areas, standard electromagnetic field areas and weak electromagnetic field areas, which can accurately identify potential high-risk areas. Subsequently, according to the communication parameters of the drone and the specific electromagnetic field strength data of each strong electromagnetic field area, the safe operating distance of the drone in each strong electromagnetic field area is calculated, so that the safety distance of the drone can be dynamically adjusted according to the actual electromagnetic environment, greatly improving the safety of the inspection. Finally, according to the three-dimensional point cloud model and the calculated safe operating distance, the initial inspection path is adjusted to obtain the final inspection path, which not only takes into account the avoidance of physical obstacles, but also fully considers the influence of the electromagnetic environment, thereby generating an optimal path that can ensure both the inspection effect and the safety of the drone. By taking the electromagnetic field strength as an important consideration, not only the communication quality and control stability of the drone in a strong electromagnetic field environment are improved, but also the risk of discharge of metal or exposed parts of the drone body is reduced, and at the same time, the overall efficiency and data quality of the inspection are improved. In addition, the dynamic path planning method based on real-time electromagnetic field strength data provides a more comprehensive safety risk assessment mechanism for substation drone inspections, effectively improving the safety and efficiency of inspections.

[0059] Other features and advantages of the embodiments of the present application will be described in detail in the subsequent specific implementation section. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 A schematic diagram of a process flow of a method for planning a substation inspection path by a drone for inspection provided in an embodiment of the present application;

[0061] Figure 2 A pseudo code for a breadth-first search provided in an embodiment of the present application;

[0062] Figure 3 A schematic diagram of a spherical safety area provided in an embodiment of the present application;

[0063] Figure 4 A schematic diagram of the structure of an inspection path planning system for a substation inspection drone provided in an embodiment of the present application. DETAILED DESCRIPTION

[0064] In order to make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. It should be understood that the specific implementation methods described herein are only used to illustrate and explain the embodiments of the present application, and are not used to limit the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0065] It should be noted that if the embodiments of the present application involve directional indications (such as up, down, left, right, front, back...), the directional indications are only used to explain the relative position relationship, movement status, etc. between the components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly.

[0066] In addition, if there are descriptions involving "first", "second", etc. in the embodiments of the present application, the descriptions of "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or suggesting their relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In addition, the technical solutions between the various embodiments can be combined with each other, but they must be based on the ability of ordinary technicians in the field to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such combination of technical solutions does not exist and is not within the scope of protection required by this application.

[0067] Example 1

[0068] Figure 1 The following is a schematic diagram of a process flow of a method for planning a substation inspection path using a drone according to an embodiment of the present application. Figure 1 As shown, an embodiment of the present application provides an inspection path planning method for a substation inspection drone, which is applied to a central control host. The central control host is connected to the drone. An electromagnetic detection device is provided on the drone. The electromagnetic detection device is used to detect the industrial frequency electric field strength and the industrial frequency magnetic field strength. The method may include the following steps.

[0069] S110, acquiring point cloud data of the substation through a drone, and constructing a three-dimensional point cloud model of the substation based on the point cloud data;

[0070] S120, in response to receiving an inspection instruction for a target inspection subject, obtaining a starting point and an end point of the UAV, and generating an initial inspection path according to the three-dimensional point cloud model, the starting point, the end point, and a preset flight altitude, wherein the target inspection subject is located on the initial inspection path;

[0071] S130, in response to receiving the power frequency electric field strength data and the power frequency magnetic field strength data of the test point of the substation transmitted by the drone, dividing the substation into m electromagnetic field areas according to the power frequency electric field strength data and the power frequency magnetic field strength data of the test point, where m is an integer greater than 1, and the m electromagnetic field areas include at least one strong electromagnetic field area, at least one standard electromagnetic field area and / or at least one weak electromagnetic field area;

[0072] S140, obtaining the communication parameters of the UAV, and determining the safe operating distance of the UAV in each strong electromagnetic field area according to the power frequency electric field strength data, power frequency magnetic field strength data and communication parameters of each strong electromagnetic field area;

[0073] S150, adjusting the initial inspection path according to the three-dimensional point cloud model and the safe operating distance of each strong electromagnetic field area to obtain a final inspection path;

[0074] S160: Send the final inspection path to the drone, so that the drone inspects the substation according to the final inspection path.

[0075] In this embodiment, the high-precision laser scanner or depth camera carried by the drone is first used to perform a full-scale scan and data collection of the substation. The laser scanner measures the three-dimensional coordinates of each point on the surface of an object by emitting a laser beam and receiving the reflected signal, while the depth camera obtains the depth information of the scene through structured light or time-of-flight method. The drone flies on a preset route to ensure coverage of the entire substation area, while maintaining a stable flight attitude to improve the accuracy of data collection.

[0076] In actual operation, the following methods can be adopted: First, design a grid-shaped flight path covering the entire substation to ensure that the drone can fully scan the substation from multiple angles. Second, set appropriate scanning resolution and overlap rate, usually the resolution can reach the centimeter level, and the overlap rate is kept above 60% to ensure the integrity and accuracy of the data. Then, perform automatic flight missions to collect point cloud data in real time during the flight.

[0077] After the acquisition is completed, the original point cloud data is imported into point cloud processing software, such as CloudCompare. First, data cleaning is performed to remove noise and outliers. Next, registration is performed to stitch the point cloud data obtained from multiple scans into a complete point cloud model. Then, noise reduction and simplification are performed to reduce the amount of data while retaining key features. Finally, classification and segmentation are performed to identify various types of equipment and structures in the substation, and finally, a three-dimensional point cloud model is obtained. This model not only contains the geometric information of the substation, but also can include attribute information such as color and reflectivity, providing an accurate spatial reference for subsequent path planning, which can accurately reflect the layout, equipment location and size of the substation, and help with obstacle avoidance and precise positioning.

[0078] After receiving the inspection instruction for the target inspection subject, the starting point and end point of the drone are obtained, and an initial inspection path is generated based on the three-dimensional point cloud model, the starting point, the end point and the preset flight altitude. The target inspection subject is located on the initial inspection path, wherein the inspection instruction for the target inspection subject is issued by the staff, and the staff can click on the equipment or building that needs to be inspected on the screen of the central control host, and the equipment or building that needs to be inspected is the target inspection subject.

[0079] Specifically, an initial inspection path can be generated based on the constructed three-dimensional point cloud model and combined with a path planning algorithm. First, obtain the take-off point and landing point of the drone. The starting point and end point of the drone can be set manually, that is, the staff can set the starting point and end point of the drone through the central control host. The starting point and end point can be a point. Then, according to the inspection instructions received, determine the target equipment or area that needs to be inspected. Next, a path planning algorithm can be used to find a path from the starting point to the end point in three-dimensional space that passes through the target inspection subject.

[0080] In specific implementation, the following method can be adopted: First, mark the take-off point, landing point and the location of the target inspection subject in the 3D point cloud model. Then, set the preset flight altitude, which can usually be set to 5 to 10 meters higher than the highest equipment in the substation. Then, discretize the 3D point cloud model into a 3D grid map, and each grid can be marked as passable or obstacle.

[0081] The preset path planning algorithm is used. The algorithm starts from the starting point, gradually explores the surrounding grids, and calculates the estimated total cost from the starting point through the current grid to the end point (including the distance traveled and the estimated remaining distance). The algorithm will give priority to exploring the grid with the smallest total cost until a path to the end point is found. In this process, the location of the target inspection subject also needs to be considered to ensure that the path passes through the target inspection subject.

[0082] In order to improve the smoothness and flyability of the path, the initial path can be optimized after it is obtained. For example, the path can be smoothed using Bezier curves or spline curves to remove sharp turns. The initial inspection path meets the following characteristics: 1) starting from the designated take-off point and arriving at the designated landing point; 2) passing through all target equipment or areas that need to be inspected; 3) maintaining a preset safe flight altitude; 4) avoiding obstacles in the substation; 5) the path is relatively smooth and meets the flight characteristics of the drone. The path generated in this way can ensure the completion of the inspection task while maximizing the flight safety of the drone.

[0083] In response to receiving the power frequency electric field strength data and the power frequency magnetic field strength data of the substation test point transmitted by the drone, the substation is divided into m electromagnetic field areas according to the power frequency electric field strength data and the power frequency magnetic field strength data of the test point, where m is an integer greater than 1, and the m electromagnetic field areas include at least one strong electromagnetic field area, at least one standard electromagnetic field area and / or at least one weak electromagnetic field area.

[0084] Specifically, based on the electromagnetic field data collected by the drone, the electromagnetic environment of the substation area is classified and divided. Power frequency electric field and power frequency magnetic field are the main sources of electromagnetic interference in the substation, and their intensity directly affects the communication quality of the drone and the working stability of the equipment. In the specific implementation, the following methods can be adopted: First, during the flight, the drone uses the electromagnetic detection device on board to continuously collect power frequency (usually 50Hz or 60Hz) electric field strength and power frequency magnetic field strength data. These data are associated with the GPS location information of the drone to form a data set containing spatial position and electromagnetic field strength. The frequency of data collection can be set to multiple times per second to ensure the continuity and integrity of the data. The electromagnetic detection device can be a power frequency electromagnetic field measuring instrument, which is used to measure the power frequency electromagnetic field strength generated by the power system and electrical equipment, and it works based on the principle of electromagnetic induction. For electric field measurement, an electric field probe is usually used; for magnetic field measurement, a magnetic field probe is used. These probes can convert the electric field or magnetic field in the environment into a measurable electrical signal.

[0085] After receiving the above electromagnetic field data, data preprocessing is first performed, including removal of outliers, smoothing, etc. Then, an interpolation algorithm (such as Kriging interpolation) is used to expand the discrete measurement point data to the entire substation area to form a continuous electromagnetic field intensity distribution map.

[0086] According to national standards or industry specifications, the threshold of electromagnetic field strength is set to divide the substation area into three types of electromagnetic field areas: strong, standard and weak. For example, areas with electric field strength greater than 10kV / m or magnetic induction strength greater than 0.5mT can be divided into strong electromagnetic field areas; areas with electric field strength between 4 and 10kV / m or magnetic induction strength between 0.1 and 0.5mT can be divided into standard electromagnetic field areas; areas below these values ​​can be divided into weak electromagnetic field areas.

[0087] The interpolated electromagnetic field intensity distribution map is segmented using an image segmentation algorithm to obtain different electromagnetic field regions. Each region is assigned a unique identifier, and its boundary coordinates and average electromagnetic field intensity value are recorded.

[0088] Finally, a detailed electromagnetic environment distribution map of the substation is obtained, which contains multiple clearly defined electromagnetic field areas. The electromagnetic environment distribution map not only reflects the spatial distribution characteristics of the electromagnetic field inside the substation, but also provides the specific intensity level of each area.

[0089] The communication parameters of the UAV are obtained, and the safe operating distance of the UAV in each strong electromagnetic field area is determined based on the industrial frequency electric field strength data, industrial frequency magnetic field strength data and communication parameters of each strong electromagnetic field area.

[0090] Specifically, based on the characteristics of the drone's communication system and the features of the strong electromagnetic field environment, the minimum distance that the drone can safely approach the strong electromagnetic field area while ensuring the communication quality is calculated.

[0091] First, it is necessary to obtain the key communication parameters of the UAV, including but not limited to: communication frequency, transmission power, receiving sensitivity, antenna gain, modulation method, coding method, etc. These parameters determine the anti-interference ability and signal transmission characteristics of the UAV communication system. Then, analyze the impact of strong electromagnetic fields on the UAV communication system. Although the power frequency electromagnetic field (50Hz or 60Hz) has a low frequency, it is very strong and may interfere with the UAV's communication system through electromagnetic induction, intermodulation, etc. This interference may cause signal distortion, increased bit error rate, communication link interruption and other problems. In actual operation, the following methods can be used to determine the safe operating distance:

[0092] First, establish a relationship model between electromagnetic field strength and distance. Ideally, the electric field strength decreases inversely with the square of the distance, and the magnetic field strength decreases inversely with the cube of the distance. Second, calculate the impact of electromagnetic interference on the communication system. This impact can be quantified using the signal-to-noise ratio (SNR) or the signal-to-interference ratio (SIR). For example, the signal-to-interference ratio can be calculated using the following formula:

[0093] SIR=P s -P i -Lp ;

[0094] Among them, Ps is the signal power transmitted by the drone (dBm), Pi is the power of the interference signal (dBm), and Lp is the path loss (dB).

[0095] Then, based on the performance indicators of the UAV communication system (such as the minimum acceptable signal-to-interference ratio), combined with the above calculation results, the minimum distance that can meet the communication quality requirements is reversely deduced. This distance is the preliminary safe operating distance.

[0096] Considering the complexity of the actual situation and potential uncertainties, this embodiment adds a safety margin based on the calculated minimum distance. For example, the actual safe operating distance can be set to 1.2 times the calculated minimum distance.

[0097] Finally, for each strong electromagnetic field area, the above calculation process is repeated, because the electromagnetic field strength in different areas may vary significantly. Finally, a set of safe operating distances for each strong electromagnetic field area is obtained. The safe operating distance can ensure the inspection effect while maximizing flight safety.

[0098] According to the 3D point cloud model and the safe working distance of each strong electromagnetic field area, the initial inspection path is adjusted to obtain the final inspection path.

[0099] Specifically, first of all, the high-precision 3D point cloud model obtained by laser scanning or photogrammetry provides a detailed spatial information basis for the entire adjustment process. This model contains key data such as the geometry, size and position of the inspection environment. At the same time, the safe working distance of each strong electromagnetic field area determined by electromagnetic field strength measurement and safety assessment provides key constraints for path adjustment.

[0100] The adjustment process begins by mapping the initial inspection path onto the 3D point cloud model. Subsequently, for each strong electromagnetic field area, a spherical or cylindrical safety zone with a safe working distance as the radius is constructed in the 3D model to visually clearly identify the dangerous areas that need to be avoided. Next, the initial path is adjusted using a path planning algorithm that searches for a feasible path in 3D space while taking into account the constraints of avoiding the safety areas. Specifically, the algorithm checks the initial path segment by segment to see if it intersects any safety area. If an intersection is found, an alternative path segment is sought that bypasses the safety area. The process involves multiple iterations until a path is found that completely avoids all safety areas.

[0101] In actual implementation, Bezier curves or spline curves can be used to smooth the path inflection points to ensure that the drone can fly smoothly along the path. In addition, a hierarchical optimization method can be used. First, a path that avoids all safety areas is quickly generated on a coarse grid, and then local optimization is performed on a finer scale based on this rough path. This ensures safety while improving the efficiency of the algorithm.

[0102] The resulting inspection path will be represented as a series of 3D coordinate points or parametric curves. These data can be directly input into the navigation system of the inspection equipment to guide the drone to complete the inspection task along a safe and efficient route. At the same time, this adjusted path can also be visualized into a 3D point cloud model, which is convenient for operators to review and make necessary manual adjustments.

[0103] This embodiment is not only applicable to static environments, but can also dynamically adjust the path based on real-time monitoring data. For example, if the electromagnetic field strength in a certain area changes, the corresponding road section can be quickly recalculated and adjusted to ensure that the inspection is always carried out within a safe range. This dynamic adaptability greatly enhances the robustness and practicality of the inspection system.

[0104] This embodiment first obtains the point cloud data of the substation through the drone and constructs a three-dimensional point cloud model, which provides an accurate spatial information basis for subsequent path planning. After receiving the inspection instruction, the initial inspection path is generated according to the starting point, end point, three-dimensional point cloud model and preset flight altitude of the drone, ensuring that the target inspection subject is located on the path, ensuring the basic coverage of the inspection. Next, the electromagnetic detection device on the drone is used to collect the power frequency electric field strength and power frequency magnetic field strength data in real time, and the substation is divided into multiple electromagnetic field areas, including strong electromagnetic field areas, standard electromagnetic field areas and weak electromagnetic field areas, which can accurately identify potential high-risk areas. Subsequently, according to the communication parameters of the drone and the specific electromagnetic field strength data of each strong electromagnetic field area, the safe operating distance of the drone in each strong electromagnetic field area is calculated, so that the safety distance of the drone can be dynamically adjusted according to the actual electromagnetic environment, greatly improving the safety of the inspection. Finally, according to the three-dimensional point cloud model and the calculated safe operating distance, the initial inspection path is adjusted to obtain the final inspection path, which not only takes into account the avoidance of physical obstacles, but also fully considers the influence of the electromagnetic environment, thereby generating an optimal path that can ensure both the inspection effect and the safety of the drone. By taking the electromagnetic field strength as an important consideration, not only the communication quality and control stability of the drone in a strong electromagnetic field environment are improved, but also the risk of discharge of metal or exposed parts of the drone body is reduced, and at the same time, the overall efficiency and data quality of the inspection are improved. In addition, the dynamic path planning method based on real-time electromagnetic field strength data provides a more comprehensive safety risk assessment mechanism for substation drone inspections, effectively improving the safety and efficiency of inspections.

[0105] In one implementation of this embodiment, generating an initial inspection path according to a three-dimensional point cloud model, a starting point, an end point, and a preset flight altitude includes the following steps:

[0106] S210, performing semantic segmentation on the three-dimensional point cloud model using a point cloud processing network to segment the three-dimensional point cloud model into n semantic regions, wherein each semantic region represents an element type;

[0107] S220, using a pattern matching algorithm to compare the n semantic regions with a pre-built standard template library of substation equipment structures, and determining the element features corresponding to each semantic region;

[0108] S230, determining the spatial relationship and functional relationship between each element in the substation according to a preset knowledge graph;

[0109] S240, obtaining a 2D design document of the substation, determining a layout rule of the substation according to the 2D design document, and determining a property of each element according to the layout rule of the substation;

[0110] S250, mapping the element type and element features corresponding to each semantic region, the spatial relationship and functional relationship between each element, and the property of each element into the three-dimensional point cloud model, so as to label each element in the three-dimensional point cloud model;

[0111] S260, mapping the target inspection subject with the elements marked in the three-dimensional point cloud model to determine the target element in the three-dimensional point cloud model;

[0112] S270, mapping the starting point, the end point, and the flight altitude to the three-dimensional point cloud model to obtain the drone starting point and the drone end point in the three-dimensional point cloud model;

[0113] S280, using a preset path planning algorithm, generates an initial inspection path in the three-dimensional point cloud model according to the drone starting point, drone end point and target element.

[0114] Point cloud processing network (hereinafter referred to as network) is a kind of deep learning technology, and its network architecture can be PointNet++ or PointCNN, which is used to process point cloud data. First, the original point cloud data is converted into a format suitable for network input, usually including features such as the three-dimensional coordinates, normal vectors and color information of the points. Then, the network extracts local and global features of the point cloud through multi-layer convolution and pooling operations. Based on the feature extraction, the network uses a fully connected layer or a deconvolution layer to classify each point into predefined semantic categories, such as transformers, switchgear, transmission lines, etc. This process needs to be iterated many times to improve the segmentation accuracy. Finally, each point is assigned a semantic label to form n semantic regions. For example, in a substation cloud model containing 1 million points, it may be segmented into 10 semantic regions, including transformer area, distribution device area, control room area, etc. Semantic segmentation can not only identify the main equipment and structures of the substation, but also distinguish environmental elements such as the ground and vegetation, providing a rich semantic information basis for subsequent path planning. Through this step, the originally disorganized point cloud data is given a clear meaning, greatly improving the pertinence and efficiency of subsequent processing.

[0115] After that, the semantic region is compared with the pre-built standard template library of substation equipment structure using a pattern matching algorithm. First, a template library containing standard 3D models of various substation equipment is constructed, and the standard 3D models include detailed geometric shapes, size ratios, and feature point information. Then, feature descriptors such as SIFT (Scale-Invariant Feature Transform) or FPFH (Fast Point Feature Histograms) are extracted for each semantic region. Next, the semantic region and the model in the template library are registered and matched using algorithms such as ICP (Iterative Closest Point). During the matching process, considering the possible deformation or partial occlusion of the actual equipment, a tolerance threshold can be set to allow a certain degree of incomplete matching. For example, for a semantic region identified as "transformer", it is first tried to match it with various transformer models in the template library, and the best matching model is found by calculating the similarity and geometric consistency of the feature points. If the matching degree exceeds the preset threshold (such as 85%), the specific equipment type and model of the semantic region is confirmed. Not only can the major categories of equipment (such as circuit breakers, disconnectors, etc.) be determined, but also specific models and specifications can be identified. Through this step, each semantic area is given more detailed and accurate element features, greatly improving the pertinence and efficiency of inspections.

[0116] The spatial and functional relationships between each element in the substation are determined according to the preset knowledge graph. Specifically, the knowledge graph is a comprehensive knowledge base that contains the topological structure, functional associations, and spatial layout rules of the substation equipment. First, a graph structure containing nodes (representing equipment) and edges (representing relationships) needs to be constructed. Nodes contain attribute information of equipment, such as type, model, rated parameters, etc.; edges represent relationships such as physical connection, electrical connection, or spatial relative position between equipment. For example, there is an electrical connection and protection relationship between the transformer and the circuit breaker, which is represented as an edge with specific attributes in the knowledge graph. Next, the knowledge graph is queried and reasoned using a graph traversal algorithm (such as depth-first search or breadth-first search). For example, given a high-voltage switch, all the devices connected to it and the functional relationships of these devices can be found through graph traversal. At the same time, the spatial layout information extracted from the point cloud data can be verified and supplemented using spatial relationship reasoning rules (such as "transformers are usually located downstream of disconnectors").

[0117] In addition, the knowledge graph also contains the logical relationships between devices, such as protection sequence, operation sequence, etc., which helps to understand the operation logic of the substation. Through this step, not only can a complete relationship network of internal devices of the substation be established, but also possible identification errors in the previous steps can be verified and corrected.

[0118] Obtain the 2D design documents of the substation and determine the layout rules and element properties based on them. First, it is necessary to parse the 2D design documents, including but not limited to CAD drawings, floor plans, and electrical system diagrams. Image processing and text recognition technology (OCR) are used in the parsing process to extract key information, such as equipment location coordinates, connection relationships, and technical parameters. This information is then compared and fused with the information extracted from the 3D point cloud model in the previous step. For example, if the 2D drawing shows that a certain location should have a specific model of circuit breaker, check whether the corresponding location in the 3D model has actually identified the device. Through this comparison, potential equipment replacements or layout changes can be discovered to ensure the timeliness of the data.

[0119] Next, based on the layout information in the 2D document, the overall layout rules of the substation are extracted, such as the minimum safe distance between equipment, the division of electrical isolation areas, etc. These rules are encoded into a series of constraints for subsequent path planning. At the same time, according to the technical specifications in the design document, the specific properties of each element are determined, such as voltage level, rated current, insulation level, etc. For example, for a 500kV main transformer, it can be marked as a high-risk area.

[0120] Element types, features, relationships, and properties are mapped to a 3D point cloud model for annotation. First, a multi-level data structure is established to store and associate various information. At the bottom layer is the original point cloud data, where each point contains its 3D coordinates and possible color information. On this basis, the results of semantic segmentation are added to divide the point cloud into different semantic regions. Then, the specific device information obtained by pattern matching is associated with these semantic regions, including device type, model, etc. Next, the spatial and functional relationships derived from the knowledge graph are added, which are expressed as connecting lines or annotations between devices. Finally, the property information extracted from the 2D document is added to the corresponding device.

[0121] For example, for a semantic area identified as a circuit breaker, its type (such as SF6 circuit breaker), model (such as a certain model of ABB), other equipment connected to it (such as adjacent disconnectors and busbars), and its voltage level, rated current and other properties will be marked. In actual applications, this kind of marking can be done in a variety of visual ways such as color coding, text labels, icons, etc., to facilitate human-computer interaction and information reading.

[0122] The target inspection subject is mapped with the elements marked in the 3D point cloud model to determine the target elements in the 3D point cloud model.

[0123] First, the inspection instructions are parsed to extract the specific objects that need to be inspected, which can be a certain type of equipment (such as all circuit breakers), a specific area (such as the main transformer area), or a specific equipment (such as main transformer No. 1). Then, search and match in the annotated 3D model. Specifically, first match the equipment type at the semantic level, then locate the specific location at the spatial level, and finally confirm the detailed information at the attribute level. For example, if the inspection target is "all circuit breakers in the 500kV distribution device area", first locate the 500kV distribution device area in the model, and then identify all elements marked as circuit breakers in the area. In this process, the importance and current status of the equipment can also be considered. For example, if a circuit breaker has a recent abnormal record, it may be given a higher inspection priority.

[0124] Map the starting point, end point and flight altitude to the 3D point cloud model. Specifically, the given starting point and end point coordinates are converted to the coordinates of the 3D point cloud model, and the spatial transformation matrix is ​​used in the conversion process to ensure accurate positioning. For example, if the given starting point is a GPS coordinate (longitude, latitude, altitude), it can be converted to XYZ coordinates in the point cloud model. Next, the mapping of the flight altitude needs to take into account the terrain and building height of the substation. A plane parallel to the ground can be created in the entire 3D model, and its height is equal to the given flight altitude plus the average height of the ground. This plane will serve as the main flight layer of the drone, and the height of the plane can also be the highest equipment in the substation plus a preset height value (such as 5m). For example, if the specified flight altitude is 20 meters, and the average ground altitude of the substation is 100 meters, the flight plane will be set at an absolute altitude of 120 meters. Finally, the determined starting point and end point coordinates are stored as input parameters for drone path planning.

[0125] Using a preset path planning algorithm, an initial inspection path is generated in the three-dimensional point cloud model according to the starting point, end point and target element of the drone. The path planning algorithm can be a heuristic search algorithm. Specifically, first, the three-dimensional point cloud model is discretized into a grid or graph structure, and each grid point or node represents a possible flight position. Then, considering the particularity of the substation environment, constraints can be set, such as a safe distance (keeping a certain distance from high-voltage equipment), a no-fly zone (such as above the control room), and a flight altitude limit. Next, the path planning algorithm starts to search for a feasible path from the starting point to the end point. In the path planning algorithm, each time the node with the smallest estimated cost is selected for expansion, where the cost function includes the distance traveled and the estimated remaining distance. In order to ensure that the path passes through all target elements, the algorithm needs to perform multiple iterations, taking the intermediate target points as sub-goals, and planning the path segment by segment. The path planning algorithm can be an A* algorithm.

[0126] The initial inspection path of this implementation not only takes the shortest distance into consideration, but also includes obstacle avoidance, which provides a reliable technical basis for UAVs to perform efficient and safe inspection tasks in complex environments, and effectively improves the feasibility and efficiency of UAV inspection tasks.

[0127] In one implementation of this embodiment, the steps of constructing a standard template library for substation equipment structure include:

[0128] S310, obtaining a site image of the substation, performing feature extraction on the site image, and obtaining geometric features and texture features of each element in the substation, the elements including buildings and equipment;

[0129] S320, using a preset edge detection algorithm and a preset corner point detection algorithm to identify the outline and key points of each element;

[0130] S330. Construct a standard template library for substation equipment structure based on the geometric features, texture features, contours and key points of each element.

[0131] After acquiring the on-site images of the substation, image preprocessing is first performed. This includes operations such as denoising, contrast enhancement, and illumination equalization. Denoising can use methods such as Gaussian filtering or median filtering to reduce random noise in the image. Contrast enhancement can use histogram equalization technology to make the grayscale distribution of the image more uniform, which is conducive to subsequent feature extraction. Illumination equalization can eliminate the uneven illumination caused by factors such as shooting time and weather conditions.

[0132] Next is image segmentation, the purpose of which is to separate different elements in the image (such as buildings, equipment). Algorithms such as threshold segmentation can be used. For each segmented element, geometric features and texture features are extracted. Geometric features include area, perimeter, aspect ratio, circularity, etc. For example, the area can be obtained by calculating the number of pixels occupied by the element, and the perimeter can be estimated by the number of edge pixels. The aspect ratio can be obtained by calculating the aspect ratio of the minimum enclosing rectangle of the element. Roundness can be calculated using 4π×area / perimeter 2 is used to distinguish circular and non-circular elements.

[0133] Texture features can be extracted using the Gray Level Co-occurrence Matrix (GLCM). It can calculate statistics such as energy, contrast, homogeneity, and entropy to describe texture. For example, energy reflects the uniformity of texture, and contrast reflects the clarity of texture.

[0134] In addition, color features such as color histograms or color moments can be extracted. They are used to distinguish different types of equipment or buildings. For example, the mean and variance of hue, saturation, and brightness in the HSV color space can be calculated.

[0135] Finally, all the extracted features are organized into feature vectors, one for each element. These feature vectors are used for subsequent element recognition and classification.

[0136] Use the preset edge detection algorithm and corner detection algorithm to identify the outline and key points of each element in the substation. Edge detection is used to identify the outline of elements. Commonly used edge detection algorithms include the Canny edge detector. Taking the Canny edge detector as an example, its workflow includes the following steps: 1. Use a Gaussian kernel to smooth the image to reduce the impact of noise. 2. Use the Sobel operator to calculate the horizontal and vertical gradients of the image, and then calculate the gradient amplitude and direction. 3. Compare the gradient amplitude along the gradient direction and retain the local maximum point. 4. Use high and low thresholds to filter edge pixels. 5. Connect the edges using the hysteresis threshold method.

[0137] Corner detection is used to identify key points of elements, which usually represent important features of the shape of an object. Commonly used corner detection algorithms include Harris corner detectors. Since corner detection algorithms are relatively common, this application will not elaborate on them.

[0138] In practical applications, multiple edge detection and corner detection algorithms can be combined to obtain more accurate and stable results. For example, the Canny algorithm can be used to obtain a preliminary edge map, and then the Hough transform can be applied to detect regular shapes such as straight lines or circles, which helps to identify columnar equipment or round insulators in substations.

[0139] For the detected contours, shape descriptors such as Fourier descriptors or shape contexts can be further extracted. These descriptors can capture the global and local features of the contours and help with subsequent template matching.

[0140] For the detected key points, local feature descriptors such as SIFT descriptors or BRIEF (Binary Robust Independent Elementary Features) descriptors can be calculated, which can be used for subsequent feature matching and object recognition.

[0141] Finally, a standard template library for substation equipment structure is constructed. First, the features extracted in the previous steps are normalized and standardized. This is because different types of features may have different numerical ranges and units. For example, the area in a geometric feature may be the number of pixels, while the energy of a texture feature such as GLCM is dimensionless. Standardization can use the Z-score method, that is, (x-μ) / σ, where x is the original value, μ is the mean, and σ is the standard deviation. Next, create a standard template for each type of substation equipment. For example, for a transformer, hundreds of images at different angles and under different lighting conditions can be collected. The features extracted from these images are statistically analyzed to calculate the mean and standard deviation of each feature. The standard template can be represented as a feature vector, in which each element is the mean of the corresponding feature, and the standard deviation is also recorded for subsequent similarity calculations.

[0142] In order to facilitate fast retrieval and matching, the template library is usually organized in a tree structure or hash structure. For example, a kd tree can be used to organize templates, making nearest neighbor search in a large-scale template library efficient. The template library also needs to contain metadata, such as the model, specification, and normal parameter range of the equipment. For example, for a specific model of lightning arrester, the template may contain its appearance characteristics when it is working normally, as well as the normal range of electrical parameters such as leakage current and resistive current.

[0143] This implementation realizes the conversion process from raw image data to a structured standard template library for substation equipment, and builds a comprehensive standard template library. The standard template library for substation equipment not only contains visual features, but also combines the technical parameters and status information of the elements, which facilitates the automated inspection and monitoring of substation equipment and effectively improves the efficiency and accuracy of power system management.

[0144] In one implementation of this embodiment, dividing the substation into m electromagnetic field areas according to the power frequency electric field strength data and the power frequency magnetic field strength data of the test point includes the following steps:

[0145] S410, using a preset interpolation algorithm, interpolating the substation according to the power frequency electric field strength data and the power frequency magnetic field strength data of all the points to be measured, to obtain an electromagnetic field strength distribution map of the substation, where the electromagnetic field strength distribution map includes multiple interpolation points and / or points to be measured;

[0146] S420, summarizing the interpolation points and / or points to be measured of the power frequency electric field strength data greater than a preset electric field strength threshold and the power frequency magnetic field strength data greater than a preset magnetic field strength threshold in the electromagnetic field strength distribution diagram to obtain m electromagnetic field regions.

[0147] Using a preset interpolation algorithm, based on the power frequency electric field strength data and power frequency magnetic field strength data of the test point in the substation, an electromagnetic field strength distribution map of the entire substation is generated. The interpolation algorithm can be a Kriging interpolation algorithm, which can consider spatial correlation and is suitable for processing spatially unevenly distributed data.

[0148] In practical applications, it is necessary to interpolate the power frequency electric field strength and the power frequency magnetic field strength respectively. For example, suppose there are 10 points to be measured in the substation, and each point has the measured value of the electric field strength E and the magnetic field strength H. Using the above kriging method, the estimated values ​​of E and H at any point in the entire substation area can be obtained.

[0149] After the interpolation is completed, the results are visualized as an electromagnetic field intensity distribution map. This can be in the form of a contour map or a heat map. The contour map can clearly show the spatial variation trend of the electromagnetic field intensity, while the heat map can intuitively show the high and low distribution of the intensity.

[0150] When generating the distribution map, a coordinate system corresponding to the actual layout of the substation is used, and an appropriate scale is selected according to the size of the substation, and a gradient color from blue to red is used to represent the change in intensity from low to high.

[0151] Through this step, the discrete measurement data are converted into a continuous spatial distribution, which not only fills the data gaps of unmeasured points but also reveals the spatial variation law of electromagnetic field intensity.

[0152] The electromagnetic field intensity distribution map generated in S410 is analyzed and processed to identify and divide areas with specific electromagnetic field intensity characteristics. First, all interpolation points and points to be tested in the electromagnetic field intensity distribution map are traversed. For each point, it is necessary to check whether its power frequency electric field intensity and power frequency magnetic field intensity exceed the preset threshold at the same time. After completing the inspection of all points, the adjacent super-threshold points are clustered into regions. This can be implemented by a connected region marking algorithm, such as an 8-connected region marking algorithm. The basic idea of ​​the algorithm is: starting from an unmarked super-threshold point, check its 8 adjacent points, and if they are also super-threshold points, give the same mark, and then recursively perform the same operation on the newly marked points until there are no new adjacent super-threshold points.

[0153] The specific implementation can use depth-first search (DFS) or breadth-first search (BFS). Taking BFS as an example, its pseudo code is as follows Figure 2 shown.

[0154] In this pseudocode, first initialize the number of markers m to 0. Traverse each unmarked super-threshold point. Create a queue Q. Add the current point (x, y) to the queue Q. When the queue Q is not empty, perform the following operations:

[0155] Take a point (i, j) from the queue Q. If the point (i, j) is not marked and is an over-threshold point, do the following:

[0156] Mark point (i, j) as m. Traverse the 8 adjacent points of point (i, j). Add the adjacent points to queue Q.

[0157] This process will divide all adjacent super-threshold points into m different regions.

[0158] In practical applications, the size and shape of the region need to be considered. A minimum area threshold can be set, and only regions exceeding this area are considered valid electromagnetic field regions. This can be achieved by counting the number of points contained in each region. At the same time,

[0159] Finally, the features of the m identified electromagnetic field regions are extracted and described, which includes calculating the area, average electric field strength, average magnetic field strength, maximum electric field strength point, maximum magnetic field strength point, etc. of each region.

[0160] This implementation method uses an interpolation algorithm to expand discrete measurement point data into a continuous electromagnetic field intensity distribution map, filling the data gap in the unmeasured area and providing a comprehensive description of the electromagnetic environment for the entire substation. On this basis, the preset threshold and spatial clustering method are used to identify and divide the electromagnetic field area with specific intensity characteristics. It can not only intuitively display the spatial distribution of the electromagnetic field, but also accurately locate the high-intensity area, realizing the conversion from point to surface and from continuous to discrete, making the analysis and management of the electromagnetic field distribution more accurate and efficient.

[0161] In one implementation of this embodiment, the communication parameters include communication signal frequency, antenna gain, bit error rate and minimum acceptable signal strength. According to the power frequency electric field strength data, power frequency magnetic field strength data and communication parameters of each strong electromagnetic field area, determining the safe operating distance of the drone in each strong electromagnetic field area includes the following steps:

[0162] S510, for any strong electromagnetic field area, input the power frequency electric field strength data, power frequency magnetic field strength data, communication signal frequency and antenna gain of each test point in the strong electromagnetic field area into a pre-built communication interference model to obtain the signal attenuation value of each test point of the UAV in the strong electromagnetic field area;

[0163] S520, determining the interference threshold of the UAV in the strong electromagnetic field area according to the maximum signal attenuation value, the bit error rate, and the minimum acceptable signal strength;

[0164] S530: Determine the safe operating distance of the UAV in the strong electromagnetic field area according to the interference threshold.

[0165] First, for each strong electromagnetic field area, obtain the power frequency electric field strength data and power frequency magnetic field strength data of all test points in the area. At the same time, obtain the key parameters of the drone communication system, including communication signal frequency and antenna gain. The communication signal frequency usually refers to the frequency used for data transmission between the drone and the ground control station, such as common frequency bands such as 2.4GHz or 5.8GHz. Antenna gain reflects the ability of the antenna to convert input power into radio waves, usually expressed in dBi. The communication interference model is built based on electromagnetic field theory and actual test data, and can simulate the impact of the electromagnetic environment on wireless communication. Through this model, the signal attenuation value of each test point of the drone in the strong electromagnetic field area can be calculated. For example, assuming that the power frequency electric field strength of a certain test point is 5kV / m, the power frequency magnetic field strength is 100μT, the communication signal frequency is 2.4GHz, and the antenna gain is 3dBi, the signal attenuation value of the point is 20dB through the communication interference model. This process needs to be carried out for all test points in the strong electromagnetic field area to comprehensively evaluate the communication environment in the area.

[0166] Next, based on the signal attenuation value calculated in the previous step, it is necessary to determine the interference threshold of the drone in the strong electromagnetic field area. In this step, in addition to using the maximum signal attenuation value, the bit error rate and the minimum acceptable signal strength also need to be considered. The bit error rate refers to the ratio of the number of received error bits to the total number of transmitted bits during data transmission, usually expressed as B (Bit Error Rate). For example, 10 -6 The B means that on average there is 1 bit error for every 1 million bits of data transmitted. The minimum acceptable signal strength refers to the lowest signal strength at which the receiving end can work normally, usually expressed in dBm. For example, for some drone communication systems, -90dBm may be a typical minimum acceptable signal strength value. The process of determining the interference threshold may be as follows: Assuming the maximum signal attenuation value is 30dB and the target bit error rate is 10 -6 , the minimum acceptable signal strength is -90dBm. Considering that communication systems usually require a certain signal-to-noise ratio to ensure reliable communication, a safety margin, such as 10dB, can be added to the minimum acceptable signal strength. Then, the interference threshold can be set to -80dBm. This means that when considering the maximum signal attenuation of 30dB, the signal strength transmitted by the drone needs to reach at least -50dBm to ensure that the receiving end can still receive a -80dBm signal in the worst case, thereby maintaining reliable communication.

[0167] Finally, according to the interference threshold value determined above, it is necessary to determine the safe operating distance of the drone in the strong electromagnetic field area. Generally, the electromagnetic field strength decreases with increasing distance. In free space, the electromagnetic field strength usually follows the inverse square law, that is, the strength decreases inversely with the square of the distance. In one embodiment, the process of determining the safe operating distance can be as follows: First, according to the transmission power and antenna characteristics of the drone communication system, the theoretical maximum communication distance under the interference threshold condition is calculated. Then, considering the particularity of the strong electromagnetic field environment, a safety factor is introduced to further reduce this distance. For example, if the theoretical maximum communication distance is 100 meters, a safety factor of 0.8 can be used to set the safe operating distance to 80 meters. In addition, the physical layout and equipment distribution of the substation need to be considered to ensure that this safe operating distance does not cause the drone to contact high-voltage equipment or other obstacles. The final safe operating distance allows the drone to be as close as possible to the equipment that needs to be inspected while ensuring communication quality and flight safety.

[0168] The safe operating distance of this embodiment can be used for subsequent path planning to ensure that the UAV always stays within a safe range when performing inspection tasks, thereby obtaining high-quality detection data and maintaining a stable and reliable communication link.

[0169] In one implementation of this embodiment, determining the interference threshold of the drone in a strong electromagnetic field area according to the maximum signal attenuation value, the bit error rate, and the minimum acceptable signal strength includes the following steps:

[0170] S610, adding the minimum receivable signal strength to the maximum signal attenuation value to obtain the minimum acceptable strength of the UAV under the maximum attenuation condition;

[0171] S620: Add the minimum acceptable intensity, the preset impact factor, and the preset safety margin to obtain a critical value of interference of the UAV in a strong electromagnetic field area, wherein the impact factor is calculated based on the bit error rate.

[0172] Signal attenuation describes the weakening of signal strength during propagation. The causes of attenuation vary, including free space loss, atmospheric absorption, multipath effects, etc. The maximum signal attenuation value is usually obtained through field testing or theoretical calculation, and represents the most severe signal attenuation that may be encountered in a given environment.

[0173] The minimum receivable signal strength is the lowest signal strength at which the drone receiver can correctly demodulate the signal. This value is directly related to the sensitivity of the receiver and is usually measured by the equipment manufacturer. It reflects the minimum signal power required by the receiver to maintain acceptable communication quality (such as a bit error rate below a certain threshold). Adding these two values ​​is actually considering the worst-case signal transmission. For example, assuming the minimum receivable signal strength is -100dBm and the maximum signal attenuation value is 80dB, then the minimum acceptable strength under maximum attenuation is -20dBm (-100dBm+80dB). This means that in order to ensure that the signal can still be received under maximum attenuation conditions, the signal strength at the transmitter needs to be at least -20dBm.

[0174] The impact factor is a parameter calculated based on the bit error rate. The bit error rate is an important indicator for measuring the performance of a digital communication system. It indicates the proportion of erroneous bits in the received bit stream. In a strong electromagnetic field environment, external interference may cause the bit error rate to increase. The impact factor is used to quantify this impact.

[0175] One way to calculate the impact factor is to use the relationship between the bit error rate and the signal-to-noise ratio. For common modulation methods, such as QPSK (quadrature phase shift keying), the relationship between the bit error rate and the signal-to-noise ratio is as follows:

[0176]

[0177] Among them, erfc is the complementary error function, BER is the bit error rate, and SNR is the signal-to-noise ratio. Through this relationship, the required signal-to-noise ratio can be inferred based on the target bit error rate, and then the impact factor can be obtained. For example, if the target bit error rate is 10 -6 , the required signal-to-noise ratio is about 10.5dB. This value can be used as a reference for the influencing factor.

[0178] The safety margin is an additional margin set to deal with unforeseen situations and measurement errors. Its size is usually determined based on experience or the requirements of specific applications, for example, it can be set to 3 to 5 dB.

[0179] Adding these three values ​​together, we get the interference threshold of the drone in the strong electromagnetic field area. The calculation formula can be expressed as:

[0180] P t =P m +F i +M s ;

[0181] Among them, Pt is the critical value of interference, Pm is the minimum acceptable intensity, Fi is the impact factor, and Ms is the safety margin.

[0182] For a specific example, assuming that the minimum acceptable strength is -20dBm (obtained from step S610), the impact factor calculated based on the bit error rate is 10dB, and the safety margin is set to 5dB, then the interference threshold is -5dBm (-20dBm+10dB+5dB).

[0183] The interference threshold represents the maximum electromagnetic field strength that a drone can withstand in a strong electromagnetic field environment. When the environmental electromagnetic field strength exceeds this value, the drone's communication system may be significantly interfered with, affecting its normal operation.

[0184] In another embodiment, the interference threshold value calculation formula is as follows:

[0185] C t =P min +M+10log 10 (-log 10 (B))+S;

[0186] Where: C t is the interference threshold (dBm), P min is the minimum acceptable signal strength (dBm), M is the maximum signal attenuation value (dB), B is the target bit error rate, and S is the safety margin (dB).

[0187] The formula is based on the minimum acceptable signal strength P min Initially, this is the minimum signal strength required for the receiver to work properly. Add the maximum signal attenuation value M to ensure that even at maximum attenuation, the signal can still reach the minimum acceptable strength. 10 (-log 10 (B) reflects the effect of the bit error rate on the required signal strength. This term is based on a variation of the Shannon-Hartley theorem and takes into account the signal-to-noise ratio required to achieve a specific bit error rate. The lower the bit error rate, the larger the value of this term, which means that a stronger signal is required to ensure communication quality. Adding a safety margin S provides additional reliability assurance.

[0188] This formula ensures that the receiving end can still obtain a sufficiently strong signal under the condition of maximum signal attenuation. By taking into account the bit error rate, the formula can automatically adjust the required signal strength according to different communication quality requirements. The lower the bit error rate, the higher the required signal strength. The introduction of the safety margin increases the robustness of the system and provides a buffer for possible unexpected situations.

[0189] Using this formula, the appropriate interference threshold can be calculated based on the specific communication system parameters and environmental conditions. This threshold can be used to guide the determination of the safe operating distance of the drone to ensure that a reliable communication link is maintained in a strong electromagnetic field environment.

[0190] For example, if P min =-90, M = 30dB, B = 10 -6 , and S = 10dB, then:

[0191] C t =-90+30+10log 10 (-log 10 (10 -6 ))+10≈-80.

[0192] This is consistent with the results described previously, confirming the validity of the formula.

[0193] According to the interference threshold, determine the safe operating distance of the drone in the strong electromagnetic field area, including the following steps:

[0194] S630, using a preset spatial path loss calculation formula and according to the interference threshold, calculating the theoretical maximum communication distance of the UAV in the strong electromagnetic field area;

[0195] S640, determining the physical layout of the substation according to the three-dimensional point cloud model, and determining the environmental constraint distance of the drone according to the physical layout;

[0196] S650: Multiply the minimum value of the theoretical maximum communication distance and the environmental constraint distance by a preset safety factor to obtain a safe operating distance of the UAV in a strong electromagnetic field area.

[0197] First, the theoretical maximum communication distance is determined using the spatial path loss calculation formula, and then the environmental constraint distance is determined based on the physical layout of the substation. Next, the physical layout of the substation is determined using a 3D point cloud model. The 3D point cloud model is composed of a large number of points, each with its spatial coordinates (x, y, z). By analyzing the distribution of these points, the location, height, and spatial relationship of various devices in the substation can be identified. For example, the specific location and size of high-voltage lines, transformers, and distribution equipment can be identified.

[0198] Based on this physical layout, the environmental constraint distance of the drone can be determined. For example, it can be set that the drone must maintain a distance of at least 10 meters from all high-voltage equipment and at least 5 meters from buildings. By analyzing the positions of various equipment and buildings in the 3D point cloud model, the minimum flight distance that meets these constraints can be calculated. This distance is the environmental constraint distance.

[0199] Finally, determine the safe operating distance of the drone in the strong electromagnetic field area. Specifically, compare the theoretical maximum communication distance and the environmental constraint distance, and select the minimum value. This is because the actual operation of the drone must meet both communication requirements and environmental safety requirements. Selecting the minimum value ensures that both conditions are met.

[0200] In this embodiment, the formula is as follows:

[0201] D s = min(k·D t , D e )·S f ;

[0202] Where: Ds is the final safe operating distance, Dt is the theoretical maximum communication distance, k is the environmental complexity coefficient (0 < k ≤ 1), De is the environmental constraint distance, and Sf is the safety factor (0 < Sf ≤ 1).

[0203] The theoretical maximum communication distance is the maximum distance calculated based on the transmission power, antenna characteristics, and interference threshold of the UAV communication system. The free space path loss model can be used for calculation:

[0204]

[0205] Where, P t is the transmission power (dBm), Ct is the interference threshold (dBm), Gt and Gr are the transmission and receiving antenna gains (dBi) respectively, and f is the signal frequency (MHz).

[0206] The environmental complexity coefficient k reflects the difference between the actual environment and the ideal free space. In a complex substation environment, due to factors such as multipath effects, reflections, and diffractions, the value of k will be less than 1. It can be determined through on-site testing or complex electromagnetic field simulations. The environmental constraint distance De is the maximum allowable distance considering the physical layout and equipment distribution of the substation. It ensures that the UAV does not come into contact with high-voltage equipment or other obstacles. The safety factor Sf is used to further reduce the operating distance. The min() function selects the smaller value between the theoretically adjusted distance k·D t and the environmental constraint distance De, ensuring that the safe operating distance does not exceed any physical limitations.

[0207] In this formula, first calculate the theoretical maximum communication distance Dt, which represents the maximum operating range under ideal conditions. Use the environmental complexity coefficient k to adjust the theoretical distance, reflecting the impact of the actual environment. Compare the adjusted distance with the environmental constraint distance De and select the smaller value to ensure that it does not exceed the physical limitations. Finally, apply the safety factor Sf to further reduce the distance and increase an additional safety margin.

[0208] For example, assume Dt = 100 meters, k = 0.9, De = 95 meters, Sf = 0.8, then: D s = min(0.9·100, 95)·0.8 = 76 meters.

[0209] In this implementation, steps S610 and S620 first consider the worst-case signal transmission to ensure that communication can be maintained under the condition of maximum signal attenuation, and further consider the impact of bit error rate and additional safety margin to obtain a more accurate critical value. Through these two steps, a clear judgment standard is provided for the safe operation of drones in strong electromagnetic field environments. This critical value can be used to evaluate whether the electromagnetic environment in a specific area is suitable for drone operations, and can also guide the design and optimization of drone communication systems to improve their anti-interference capabilities in strong electromagnetic field environments.

[0210] The two steps S630 and S640 take into account multiple key factors: communication capability, environmental constraints, and safety margin. Through theoretical calculations and actual environmental analysis, the theoretical maximum communication distance and environmental constraint distance are obtained respectively. These two distances limit the operating range of the drone from different angles. Combining these two distances and introducing a safety factor, a conservative and reliable safe operating distance is finally obtained. Not only the theoretical communication limitations, but also the physical constraints of the actual environment are taken into account, and additional safety guarantees are added by introducing a safety factor. The obtained safe operating distance can effectively balance operational efficiency and safety, and provide reliable operational guidance for the application of drones in strong electromagnetic field environments such as substations, significantly improving the safety of drone operations, reducing the risk of accidents, while ensuring communication quality and improving operational efficiency.

[0211] In one implementation of this embodiment, the initial inspection path is adjusted according to the three-dimensional point cloud model and the safe working distance of each strong electromagnetic field area to obtain the final inspection path, including the following steps:

[0212] S710, dividing the initial inspection path into multiple path segments according to the m electromagnetic field areas, and determining a target path segment in a strong electromagnetic field area among the multiple path segments;

[0213] S720, for each target path segment, construct a spherical safety area with a safe working distance as a radius in the three-dimensional point cloud model, and the center point of the target path segment is located at the center of the spherical safety area;

[0214] S730, determining whether the initial inspection path intersects with p spherical safety areas, where p is a positive integer;

[0215] S740, if there is an initial inspection path that intersects with p spherical safety areas, determine a trajectory point on the initial inspection path that is within the spherical safety area;

[0216] S750: For any trajectory point, determine the replacement point corresponding to it, and convert all trajectory points into the corresponding replacement points to obtain a final inspection path.

[0217] First, according to the three-dimensional point cloud model and the safe working distance of each strong electromagnetic field area, the initial inspection path is adjusted to obtain the final inspection path. The first step is to divide the initial inspection path into multiple path segments according to the m electromagnetic field areas, and determine the target path segments in the strong electromagnetic field area in these path segments. In specific implementation, the spatial analysis algorithm can be used to determine the positional relationship between the initial inspection path and each electromagnetic field area. For example, the ray projection method or the polygon inclusion judgment algorithm can be used to determine which parts of the path are located in the electromagnetic field area. Suppose there is an initial inspection path with a length of 100 meters and passes through 3 electromagnetic field areas along the way. Through spatial analysis, it may be found that the three sections of the path from 20 to 30 meters, 55 to 65 meters, and 80 to 90 meters are located in these electromagnetic field areas. These path segments are marked as target path segments.

[0218] In the second step, for each target path segment, a spherical safety area with a safe working distance as a radius is constructed in the three-dimensional point cloud model, and the center point of the target path segment is located at the center of the spherical safety area. Figure 3 A schematic diagram of a spherical safety area provided in an embodiment of the present application is shown. Figure 3 As shown, AB is the target path segment, O is the center point of the target path point, and at the same time, O is the center of the spherical safety area, and OA and OB are both safe working distances. This embodiment uses three-dimensional geometric modeling technology to create a virtual safety area in the point cloud model. During specific implementation, it is first necessary to calculate the coordinates of the center point of each target path segment. Taking the 20 to 30 meter path segment identified in the previous step as an example, the coordinates of its center point may be (25,0,0). Assuming that the safe working distance of the area is 5 meters, a spherical safety area is constructed with (25,0,0) as the center and 5 meters as the radius. Three-dimensional modeling functions, such as the createSphere() function, can be used.

[0219] The third step is to determine whether the initial inspection path intersects with p spherical safety areas, where p is a positive integer. The line-surface intersection judgment algorithm in computational geometry can be used. In specific implementation, the initial inspection path can be regarded as a series of line segments, and then these line segments are judged one by one whether they intersect with each spherical safety area. A common method is to first perform a fast bounding box test. If the bounding box of the line segment does not intersect with the bounding box of the sphere, it can be directly determined that they do not intersect; otherwise, the shortest distance from the line segment to the center of the sphere is further calculated. If this distance is less than the radius of the sphere, it is determined to intersect. For example, for the spherical safety area mentioned above with (25,0,0) as the center and a radius of 5 meters, each line segment in the initial path can be checked to see if it intersects with the sphere. Suppose there is a line segment from (20,0,0) to (30,0,0). Obviously, this line segment intersects with the sphere.

[0220] The fourth step is to determine the trajectory points on the initial inspection path that are located in the spherical safety area when it is determined that there are intersections between the initial inspection path and p spherical safety areas. This can be based on the calculation and comparison of the distance from the point to the center of the sphere. In specific implementation, sampling points can be taken at fixed intervals along the initial inspection path, and then it is determined whether each sampling point is located in any spherical safety area. The judgment method is to calculate the distance from the sampling point to the center of the sphere. If this distance is less than the radius of the sphere, the point is located inside the sphere. For example, for the spherical safety area mentioned above, a sampling point can be taken every 0.1 meter along the path segment of 20 to 30 meters. For the sampling point (23,0,0), the distance from the sampling point to the center of the sphere (25,0,0) is calculated to be 2 meters, which is less than the radius of the sphere of 5 meters, so this point is marked as a trajectory point that needs to be adjusted. In this way, all trajectory points that need to be adjusted can be accurately identified, providing a basis for subsequent path optimization. This fine-grained analysis can ensure the accuracy of path adjustment and avoid unnecessary over-adjustment.

[0221] In the last step, for any trajectory point, determine the candidate point corresponding to it, and convert all trajectory points into corresponding candidate points, so as to obtain the final inspection path. It can be implemented based on the point migration algorithm of the safety distance constraint. In the specific implementation, the spherical uniform sampling algorithm can be used to find suitable candidate points. In one embodiment, for each trajectory point that needs to be adjusted, multiple points are uniformly sampled on a sphere with the point as the center and the safe working distance as the radius, and then the point closest to the original path and not in any safe area is selected as a candidate point. For example, for the trajectory point located at (23,0,0), 100 points can be sampled on a sphere with this point as the center and a radius of 5 meters. Assuming that there is a point (27,3,4) that meets the safety distance requirements and is the closest to the original path among all the sampled points, then this point is selected as a candidate point. Repeat this process for all trajectory points that need to be adjusted, and finally obtain a series of candidate points, which are connected to form the final inspection path. This method can keep the path as smooth and close to the original path as possible while ensuring safety.

[0222] This implementation can effectively identify and avoid areas with strong electromagnetic fields, ensuring the safety of inspection operations. Through precise spatial analysis and geometric calculations, the path adjustment can be minimized, and the characteristics of the original path can be maintained as much as possible while ensuring safety. Due to the use of a three-dimensional point cloud model, it can adapt to complex terrain and environments. The inspection path finally generated not only meets safety requirements, but also maintains a high inspection efficiency, providing a reliable path planning solution for actual inspection operations.

[0223] In one implementation of this embodiment, for any trajectory point, determining the replacement point corresponding thereto includes the following steps:

[0224] S810, for any trajectory point, uniformly sample multiple candidate points on the spherical surface of the spherical safety area;

[0225] S820: Select, from among multiple candidate points, a point that is closest to the trajectory point and is not within any spherical safety area as a replacement point corresponding to the trajectory point.

[0226] A spherical safety zone is a spherical space with a certain radius centered at a specific point. In drone path planning, a spherical safety zone represents obstacles or dangerous areas that need to be avoided. Track points are points on the original flight path of the drone. If these points fall within the spherical safety zone, alternative points need to be found to ensure flight safety.

[0227] In one embodiment, a spherical coordinate system can be used to generate uniformly distributed points by controlling the zenith angle θ and the azimuth angle φ. The specific implementation is as follows:

[0228] 1. Determine the number of sampling points N. Usually 100 to 1000 points can be selected. 2. Generate sampling points using the spherical coordinate formula. For each sampling point, its coordinates can be expressed as:

[0229]

[0230] z = R × cos(θ);

[0231] where R is the radius of the sphere, θ is the zenith angle (0 to π), and φ is the azimuth angle (0 to 2π).

[0232] 3. To achieve uniform distribution, θ and φ can be generated using the following method:

[0233] θ=arccos(1-2×i / N),i:0→N-1

[0234]

[0235] Select the most suitable replacement point from multiple candidate points. First, two main checks need to be performed for each candidate point: 1. Calculate the distance from the candidate point to the original trajectory point. 2. Determine whether the candidate point is located in any other spherical safety area.

[0236] Distance calculation can use the Euclidean distance formula. For two points P1(x1,y1,z1) and P2(x2,y2,z2) in three-dimensional space, the distance D between them can be calculated by the following formula:

[0237]

[0238] Next, we need to check whether each candidate point falls within the other spherical safe regions. We can do this by calculating the distance from the candidate point to the center of each safe region and comparing it with the radius of that safe region. If the distance is less than the radius, the point is within the safe region and should be excluded.

[0239] For example, suppose there is another spherical safety zone with a center coordinate of (10,10,10) and a radius of 5 meters. For the above candidate point (12,8,7), calculate its distance to the center of this safety zone:

[0240] D=√[(12-10) 2 +(8-10) 2 +(7-10) 2 ]=√(4+4+9)=√17≈4.12m;

[0241] Since 4.12 meters is smaller than the safety zone radius of 5 meters, this candidate point is within the safety zone and cannot be used as a valid replacement point.

[0242] After completing the distance calculation and safety check of all candidate points, you need to select the point closest to the original trajectory point from the remaining valid points as the replacement point. This can be achieved by comparing the distance values ​​of all valid points and selecting the point corresponding to the minimum value.

[0243] The specific steps are as follows: 1. Initialize a variable to store the minimum distance, and set the initial value to infinity. 2. Initialize a variable to store the best replacement point, and set the initial value to null. 3. Traverse each candidate point: a. Calculate the distance from the candidate point to the original trajectory point. b. Check whether the candidate point is in any other safe area. c. If it is not in any safe area and the distance is less than the current minimum distance, update the minimum distance and the best replacement point. 4. After the traversal is completed, the best replacement point is the required replacement point.

[0244] In summary, an optimal replacement position can be found for each trajectory point that needs to be replaced. The replacement position meets two key conditions: it is not within any spherical safety area, and it is the closest to the original trajectory point among all candidate points. This ensures flight safety while maintaining the characteristics of the original trajectory to the greatest extent possible.

[0245] For each trajectory point that needs to be replaced, this embodiment has a set of candidate points distributed on the surface of the spherical safety area. These candidate points provide the basis for selecting the best replacement point in the next step. The uniform distribution characteristic ensures that there are diverse choices when selecting replacement points, increasing the possibility of finding suitable replacement points. In addition, it also ensures that the replacement point is always located on the boundary of the safety area, neither entering the dangerous area nor too far away from the original trajectory point. By uniformly sampling on the surface of the spherical safety area, a set of widely distributed candidate points is provided, and the best replacement point is screened out from the candidate points through distance calculation and safety inspection. It takes into account both safety (ensuring that the replacement point is not in any dangerous area) and the continuity and smoothness of the trajectory (selecting the replacement point closest to the origin). The new trajectory obtained in this way can effectively avoid all dangerous areas while maintaining the characteristics of the original trajectory as much as possible. It significantly improves the safety of drone flight while maintaining flight efficiency, and is suitable for path planning and dynamic obstacle avoidance in complex environments.

[0246] Example 2

[0247] Based on Example 1, after the final inspection path is obtained, the inspection blind area of ​​the final inspection path can be identified, specifically including:

[0248] First, the entire inspection area is discretized into a three-dimensional grid, and each grid cell represents a potential inspection point. Then, a simulated inspection is performed along the final inspection path to calculate the visibility and reachability of each grid cell. This can be achieved using a ray tracing algorithm. A virtual ray is emitted from each point on the inspection path to the surrounding grid cells. If the ray can reach the grid cell and is not blocked by obstacles, the grid cell is considered visible.

[0249] In specific implementation, a visibility threshold can be set. For example, if the visible time of a grid unit during the entire inspection process is less than 10% of the total inspection time, it will be marked as a potential inspection blind spot. At the same time, the performance parameters of the inspection equipment, such as the camera's field of view and effective detection distance, need to be considered. For example, assuming that the camera used for inspection has a horizontal field of view of 60 degrees and an effective detection distance of 30 meters, these parameters need to be included in the calculation when performing visibility analysis.

[0250] In addition, spatial interpolation techniques can be used to evaluate the continuity of inspection coverage. For example, the sampling points on the inspection path are interpolated using the Kriging interpolation method to generate a continuous coverage intensity field. By analyzing the distribution of this intensity field, areas with weak coverage can be identified, which may be potential inspection blind spots.

[0251] For the identified inspection blind spots, the following measures can be taken: 1. Adjust the inspection path to get closer to or pass through the blind spot to improve coverage. The ant colony algorithm can be used to find the optimal path. 2. Add static observation points or hovering points at key locations to supplement the monitoring of blind spots. 3. When passing near the blind spot, design multiple observation angles and postures to maximize coverage.

[0252] For the unavoidable problem of safe working distance, the following strategies can be adopted:

[0253] 1. Use remote sensing technologies such as telephoto lenses, high-resolution sensors, or thermal imaging devices to improve detection capabilities while maintaining a safe distance. For example, if a conventional camera cannot obtain a sufficiently clear image at a safe distance of 5 meters, consider using a camera with a 30x optical zoom capability, so that even at a distance of 15 meters, you can obtain image clarity equivalent to a distance of 5 meters. 2. Combine different types of sensors, such as visible light cameras, infrared cameras, and lidar, to obtain more comprehensive information. By fusing data from multiple sensors, the shortcomings of a single sensor in long-distance observation can be compensated to a certain extent. 3. Dynamically adjust the sampling frequency and accuracy based on real-time acquired data. When approaching but not reaching the safe distance boundary, increase the frequency and accuracy of data acquisition; when far away from the dangerous area, appropriately reduce the sampling density to save resources. 4.4 Based on the acquired data, build a virtual 3D model of the inspection area. Operators can conduct detailed inspections in a safe virtual environment to make up for the shortcomings of close observation due to safety distance restrictions in actual inspections.

[0254] Through the comprehensive application of strategies, inspection blind spots can be effectively identified and handled, and when encountering unavoidable safety operating distance restrictions, the inspection quality and coverage can still be guaranteed to the greatest extent. This multi-level and multi-angle inspection strategy not only improves the comprehensiveness and accuracy of inspections, but also ensures the safety of inspection personnel and equipment, providing more reliable technical support for the maintenance and management of power facilities.

[0255] The embodiment of the present application also provides a central control host, including:

[0256] a memory configured to store instructions; and

[0257] The processor is configured to call instructions from the memory and implement the above-mentioned inspection path planning method for the substation inspection drone when executing the instructions.

[0258] Reference Figure 4 The present application also provides a patrol path planning system for a substation patrol drone, including:

[0259] Central control host;

[0260] The drone is connected to the central control host, and is provided with an electromagnetic detection device for detecting the strength of the power frequency electric field and the power frequency magnetic field.

[0261] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0262] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram and the combination of the processes and / or blocks in the flowchart and / or block diagram can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0263] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0264] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0265] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0266] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0267] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0268] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0269] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.

Claims

1. A method for planning inspection paths for substation inspection drones, characterized in that: Applied to a central control host, the central control host is connected to a drone, the drone is provided with an electromagnetic detection device, the electromagnetic detection device is used to detect the power frequency electric field strength and the power frequency magnetic field strength, the method includes: Acquiring point cloud data of the substation through the drone, and constructing a three-dimensional point cloud model of the substation based on the point cloud data; In response to receiving an inspection instruction for a target inspection subject, obtaining a starting point and an end point of the drone, and generating an initial inspection path according to the three-dimensional point cloud model, the starting point, the end point and a preset flight altitude, wherein the target inspection subject is located on the initial inspection path; In response to receiving the power frequency electric field strength data and the power frequency magnetic field strength data of the test point of the substation transmitted by the drone, the substation is divided into m electromagnetic field areas according to the power frequency electric field strength data and the power frequency magnetic field strength data of the test point, where m is an integer greater than 1, and the m electromagnetic field areas include at least one strong electromagnetic field area, at least one standard electromagnetic field area and / or at least one weak electromagnetic field area; Acquire the communication parameters of the drone, and determine the safe operating distance of the drone in each of the strong electromagnetic field areas according to the power frequency electric field strength data, the power frequency magnetic field strength data and the communication parameters of each of the strong electromagnetic field areas; According to the three-dimensional point cloud model and the safe operating distance of each of the strong electromagnetic field areas, the initial inspection path is adjusted to obtain a final inspection path; The final inspection path is sent to the drone, so that the drone inspects the substation according to the final inspection path.

2. The method according to claim 1, characterized in that The generating of the initial inspection path according to the three-dimensional point cloud model, the starting point, the end point and the preset flight height includes: Performing semantic segmentation on the three-dimensional point cloud model using a point cloud processing network to segment the three-dimensional point cloud model into n semantic regions, wherein each of the semantic regions represents an element type; Using a pattern matching algorithm, the n semantic regions are compared with a pre-built standard template library of substation equipment structures to determine the element features corresponding to each semantic region; Determine the spatial relationship and functional relationship between each of the elements in the substation according to a preset knowledge graph; Acquire a 2D design document of the substation, determine a layout rule of the substation according to the 2D design document, and determine a property of each of the elements according to the layout rule of the substation; Mapping the element type and element features corresponding to each semantic region, the spatial relationship and functional relationship between each element, and the property of each element into the three-dimensional point cloud model to annotate each element in the three-dimensional point cloud model; Mapping the target inspection subject with the elements marked in the three-dimensional point cloud model to determine the target element in the three-dimensional point cloud model; Mapping the starting point, the end point and the flight altitude to the three-dimensional point cloud model to obtain a drone starting point and a drone end point in the three-dimensional point cloud model; A preset path planning algorithm is used to generate an initial inspection path in the three-dimensional point cloud model according to the drone starting point, the drone end point and the target element.

3. The method according to claim 2, characterized in that The steps of constructing the substation equipment structure standard template library include: Acquire a site image of the substation, perform feature extraction on the site image, and obtain geometric features and texture features of each element in the substation; Using a preset edge detection algorithm and a preset corner point detection algorithm to identify the outline and key points of each of the elements; A standard template library of substation equipment structure is constructed according to the geometric features, texture features, contours and key points of each element.

4. The method according to claim 2, characterized in that: The step of dividing the substation into m electromagnetic field areas according to the power frequency electric field strength data and the power frequency magnetic field strength data of the point to be measured comprises: Using a preset interpolation algorithm, interpolate the substation according to the power frequency electric field strength data and the power frequency magnetic field strength data of all the points to be measured, to obtain an electromagnetic field strength distribution map of the substation, wherein the electromagnetic field strength distribution map includes a plurality of interpolation points and / or points to be measured; The interpolation points and / or points to be measured of the power frequency electric field strength data greater than a preset electric field strength threshold and the power frequency magnetic field strength data greater than a preset magnetic field strength threshold in the electromagnetic field strength distribution diagram are summarized to obtain m electromagnetic field regions.

5. The method according to claim 1, characterized in that: The communication parameters include communication signal frequency, antenna gain, bit error rate and minimum acceptable signal strength. The safe operating distance of the UAV in each strong electromagnetic field area is determined according to the power frequency electric field strength data, the power frequency magnetic field strength data and the communication parameters of each strong electromagnetic field area, including: For any of the strong electromagnetic field areas, the power frequency electric field strength data, the power frequency magnetic field strength data, the communication signal frequency and the antenna gain of each test point in the strong electromagnetic field area are input into a pre-constructed communication interference model to obtain the signal attenuation value of each test point of the drone in the strong electromagnetic field area; Determine the interference threshold of the UAV in the strong electromagnetic field area according to the maximum signal attenuation value, the bit error rate and the minimum acceptable signal strength; The safe operating distance of the UAV in the strong electromagnetic field area is determined according to the interference threshold.

6. The method according to claim 5, characterized in that Determining the interference threshold of the UAV in the strong electromagnetic field area according to the maximum signal attenuation value, the bit error rate and the minimum acceptable signal strength includes: Adding the minimum receivable signal strength to the maximum signal attenuation value to obtain the minimum acceptable strength of the UAV under the maximum attenuation condition; Adding the minimum acceptable intensity, a preset impact factor, and a preset safety margin to obtain a critical value of interference of the UAV in the strong electromagnetic field area, wherein the impact factor is calculated based on the bit error rate; Determining the safe operating distance of the UAV in the strong electromagnetic field area according to the interference threshold value includes: Using a preset spatial path loss calculation formula, according to the interference critical value, the theoretical maximum communication distance of the UAV in the strong electromagnetic field area is calculated; Determining the physical layout of the substation according to the three-dimensional point cloud model, and determining the environmental constraint distance of the drone according to the physical layout; The minimum value of the theoretical maximum communication distance and the environmental constraint distance is multiplied by a preset safety factor to obtain a safe operating distance of the UAV in the strong electromagnetic field area.

7. The method according to claim 1, characterized in that The step of adjusting the initial inspection path according to the three-dimensional point cloud model and the safe operating distance of each of the strong electromagnetic field areas to obtain a final inspection path includes: Dividing the initial inspection path into a plurality of path segments according to the m electromagnetic field regions, and determining a target path segment in the strong electromagnetic field region among the plurality of path segments; For each of the target path segments, construct a spherical safety area with the safe working distance as a radius in the three-dimensional point cloud model, wherein the center point of the target path segment is located at the center of the spherical safety area; Determine whether the initial inspection path intersects with p spherical safety areas, where p is a positive integer; If there is an intersection between the initial inspection path and p spherical safety areas, then determining a trajectory point on the initial inspection path that is within the spherical safety area; For any of the track points, a corresponding replacement point is determined, and all of the track points are converted into the corresponding replacement points to obtain a final inspection path.

8. The method according to claim 7, characterized in that The step of determining a replacement point corresponding to any one of the trajectory points includes: For any of the trajectory points, multiple candidate points are uniformly sampled on the spherical surface of the spherical safety area; Among the plurality of candidate points, a point which is closest to the trajectory point and is not within any of the spherical safety regions is selected as a replacement point corresponding to the trajectory point.

9. A central control host, characterized in that: include: a memory configured to store instructions; as well as The processor is configured to call the instruction from the memory and to implement the inspection path planning method of the substation inspection drone according to any one of claims 1 to 8 when executing the instruction.

10. A patrol path planning system for a substation inspection drone, characterized in that: include: The central control host according to claim 9; The drone is connected to the central control host, and the drone is provided with an electromagnetic detection device for detecting the strength of the power frequency electric field and the strength of the power frequency magnetic field.

Images