All-terrain photovoltaic power station intelligent weeding robot control system based on shielding real-time monitoring

Abstract

The invention discloses an all-terrain photovoltaic power station intelligent weeding robot control system based on shielding real-time monitoring, and relates to the technical field of intelligent planning. Environmental data of a photovoltaic power station are collected, an environmental data set S is constructed and preprocessed, a photovoltaic power station three-dimensional model is constructed, and the three-dimensional model of the photovoltaic power station is constructed according to the three-dimensional model of the photovoltaic power station; using a path planning algorithm to obtain an operation path, collecting operation data of the intelligent weeding robot in real time in the weeding operation process, constructing an operation data set D, and obtaining a path comprehensive score LJ of each grid region according to the operation data set D, and comparing and analyzing the path comprehensive score LJ with a preset first path score threshold LY and a preset second path score threshold LP, evaluating the operation path, and carrying out operation path adjustment on the grid region of which the evaluation result is that the operation path has deviation, so that the instantaneity and the accuracy of the weeding operation are improved, and the weeding operation efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent planning technology, and in particular to an all-terrain photovoltaic power station intelligent weeding robot control system based on real-time monitoring of occlusion. Background Art

[0002] With the continuous development of intelligent technology, robot applications have penetrated into the fields of agriculture, environmental protection and energy. In the field of new energy, especially in the maintenance and management of photovoltaic power stations, how to effectively remove weeds around photovoltaic panels has become an important issue in improving the efficiency of photovoltaic power stations. Photovoltaic power stations are generally located in relatively open areas and are usually in direct contact with the natural environment. Therefore, the growth of weeds not only affects the operating efficiency of photovoltaic panels, but may also cause damage to equipment. In order to improve work efficiency and reduce labor costs, more and more photovoltaic power stations have begun to use intelligent weeding robots to perform ground cleaning work. In this application scenario, the robot not only needs to perform weeding operations efficiently and accurately, but also needs to be able to avoid obstacles on different terrains to ensure the continuity and safety of the work process.

[0003] Although existing intelligent weeding robots can complete simple weeding tasks to a certain extent, they often perform poorly when faced with complex terrain and dynamic environments. Especially in terms of obstacle avoidance and path planning, existing intelligent robot systems often find it difficult to effectively deal with rapidly changing obstacles, such as fallen leaves, stones, and temporarily placed equipment. Traditional robots mainly rely on simple sensors to identify obstacles in front of them, but these sensors often find it difficult to update in real time and accurately determine the location and characteristics of obstacles, making it difficult for robots to flexibly adjust their paths in complex environments. Especially in all-terrain operations, the robot's obstacle avoidance capabilities and the real-time performance of path planning are greatly challenged. In addition, existing technologies are also difficult to accurately predict and respond to dynamically changing obstacles, causing robots to frequently stop and reposition, and even leading to reduced work efficiency and equipment damage. Summary of the invention

[0004] In view of the deficiencies in the prior art, the present invention provides an all-terrain photovoltaic power station intelligent weeding robot control system based on real-time monitoring of occlusion, which solves the problems in the above-mentioned background technology.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: an all-terrain photovoltaic power station intelligent weeding robot control system based on real-time monitoring of occlusion, including a data acquisition module, a three-dimensional model building module, a path planning module, an operation path evaluation module and an operation path optimization module;

[0006] The data acquisition module is used to install a laser radar device and a visual sensor in the photovoltaic power station, and collect point cloud data and image data of the photovoltaic power station based on the laser radar device and the visual sensor to construct an environmental data set S;

[0007] The three-dimensional model building module is used to perform preprocessing according to the environmental data set S, and to construct a three-dimensional model of the photovoltaic power station according to the preprocessed environmental data set S, and to divide the three-dimensional model of the photovoltaic power station into a plurality of grid areas;

[0008] The path planning module is used to obtain the operation path of the intelligent weeding robot based on the three-dimensional model of the photovoltaic power station using a path planning algorithm, and perform the weeding task according to the operation path;

[0009] The operation path evaluation module is used to deploy laser radar equipment, ultrasonic sensors and high-definition cameras on the intelligent weeding robot, collect the operation data of the intelligent weeding robot in real time, and construct an operation data set D. According to the operation data set D, the path comprehensive score LJ of each grid area is obtained, and the first path score threshold LY and the second path score threshold LP are preset. The first path score threshold LY and the second path score threshold LP are compared and analyzed with the path comprehensive score LJ to evaluate the operation path and obtain the grid area where the operation path has deviations;

[0010] The operation path optimization module is used to obtain the path adjustment amount ΔP according to the operation data set D for the grid area where the evaluation result shows that the operation path has deviation, and generate a control instruction according to the path adjustment amount ΔP to control the intelligent weeding robot to adjust the operation path.

[0011] Preferably, the data acquisition module includes a point cloud data acquisition unit and an image data acquisition unit;

[0012] The point cloud data acquisition unit is used to install a laser radar device in the photovoltaic power station, and use the laser radar device to monitor the photovoltaic power station in real time, obtain point cloud data of the photovoltaic power station, and construct a point cloud data group, wherein the point cloud data group includes the terrain, obstacles and point cloud data of the photovoltaic power station and photovoltaic panels;

[0013] The image data acquisition unit is used to install a visual sensor in the photovoltaic power station, and use the visual sensor to collect image data of the photovoltaic power station to construct an image data group;

[0014] The image data set includes shadow area, obstacles and weed cover image data of the photovoltaic power station;

[0015] The constructed point cloud data set and image data set are transmitted to the central database of the photovoltaic power station in real time through the wireless transmission equipment carried by the laser radar equipment and the visual sensor, and the environmental data set S is constructed.

[0016] Preferably, the three-dimensional model building module includes a data preprocessing unit and a three-dimensional model building unit;

[0017] The data preprocessing unit is used to perform point cloud data preprocessing and image data preprocessing according to the constructed environment data set S;

[0018] The point cloud data preprocessing refers to removing isolated points in the point cloud data group in the environmental data set S using a statistical outlier removal method, simplifying the point cloud data using a voxel grid method, and performing point cloud region segmentation using a region growing algorithm;

[0019] The image data preprocessing is used to remove image noise using Gaussian filtering, enhance shadow area detail information using a contrast stretching algorithm, and segment vegetation coverage area images and obstacle images using a deep learning algorithm for the image data group in the environmental data set S;

[0020] The three-dimensional model building unit is used to build a three-dimensional model of the photovoltaic power station using 3D modeling software based on the pre-processed environmental data set S, and divide the three-dimensional model of the photovoltaic power station into a plurality of grid areas, each grid area represents a weeding task point;

[0021] The photovoltaic power station 3D model is based on the point cloud data set of the photovoltaic power station, and the photovoltaic power station 3D model framework is constructed. Feature data is extracted based on the image data set of the photovoltaic power station, and the extracted feature data is mapped to the photovoltaic power station 3D model framework to obtain a complete photovoltaic power station 3D model.

[0022] The three-dimensional model of the photovoltaic power station includes the layout of photovoltaic panels, the location and height of obstacles, and the terrain changes of the photovoltaic power station.

[0023] Preferably, the path planning module is used to perform a grid search in the three-dimensional model of the photovoltaic power station using a path planning algorithm based on the three-dimensional model of the photovoltaic power station, and to perform obstacle avoidance operation path planning based on the positions of obstacles in the three-dimensional model of the photovoltaic power station, to obtain an operation path from the starting position of the intelligent weeding robot to the end point of the weeding operation task, wherein the operation path includes all weeding operation task points of the intelligent weeding robot and the coordinates of the weeding operation task points of the intelligent weeding robot, and the operation path is sent to the intelligent weeding robot to perform the weeding operation task.

[0024] Preferably, the operation path evaluation module includes an operation data acquisition unit, a data analysis unit and an evaluation unit;

[0025] The operation data collection unit is used to deploy laser radar equipment, ultrasonic sensors, high-definition cameras and high-precision GPS on the intelligent weeding robot. During the weeding operation, the horizontal distance D between the intelligent weeding robot and the obstacle, the real-time weeding task point (x act ,y act ), weed height H, weed density ρ, weed area R, lateral projection area S of the obstacle and height H of the obstacle, and install a light sensor on the photovoltaic panel to collect the angle θ between the light direction and the obstacle, and based on the collected horizontal distance D between the intelligent weeding robot and the obstacle, the real-time weeding task point (x act ,y act ), weed height H, weed density ρ, weed area R, lateral projection area S of the obstacle, obstacle height H and angle θ between the light direction and the obstacle, to construct the operation data set D;

[0026] The data analysis unit is used to perform preprocessing according to the operation data set D, including denoising, outlier processing and data standardization, and perform summary calculation according to the preprocessed operation data set D to obtain the obstacle shielding coefficient ZD, weed influence coefficient ZC and path deviation coefficient PX of each grid area;

[0027] The obstacle shielding coefficient ZD is obtained in the following way:

[0028]

[0029] In the formula, H i represents the height of the ith obstacle, D i represents the horizontal distance from the position of the intelligent weeding robot to the i-th obstacle, S i represents the lateral projection area of ​​the i-th obstacle, θ i represents the angle between the illumination direction and the i-th obstacle, n represents the total number of obstacles in the grid area, i = 1, 2, 3, ..., n;

[0030] The weed influence coefficient ZC is obtained as follows:

[0031]

[0032] In the formula, H g represents the average height of weeds in the grid area, ρ g represents the weed density in the grid area, R w is the weed area in the grid area, α and β represent the average height of weeds in the grid area, H g and the weed density ρ within the grid area gThe weight coefficient is 0<α<1, 0<β<1, and α+β=1.

[0033] Preferably, the path deviation coefficient PX is obtained in the following manner:

[0034]

[0035] In the formula, x act (t) represents the x-axis coordinate of the intelligent weeding robot at the operation time point t, y act (t) represents the y-axis coordinate of the intelligent weeding robot at the operation time point t, x pre (t) represents the x-axis coordinate of the preset path of the intelligent weeding robot at the operation time point t, y pre (t) represents the y-axis coordinate of the preset path of the intelligent weeding robot at the operation time point t, represents the real-time weeding task point of the intelligent weeding robot (x act ,y act ) and the preset weeding task point (x pre ,y pre ), T represents the total operation time of the intelligent weeding robot in the grid area; t represents the t-th operation time point of the intelligent weeding robot in the grid area, t = 1, 2, 3, ..., T; wherein the preset operation point (x pre ,y pre ) is obtained through the planned operation path.

[0036] Preferably, the evaluation unit is used to perform summary calculation based on the obstacle shielding coefficient ZD, the weed influence coefficient ZC and the path deviation coefficient PX to obtain the path comprehensive score LJ, and the path comprehensive score LJ is obtained in the following manner:

[0037]

[0038] Wherein, δ1, δ2 and δ3 represent the weight coefficients of the obstacle shielding coefficient ZD, the weed influence coefficient ZC and the path deviation coefficient PX, respectively, 0<δ1<1, 0<δ2<1, 0<δ3<1, and δ1+δ2+δ3=1, C represents the correction constant.

[0039] Preferably, a first path scoring threshold LY and a second path scoring threshold LP are preset, and the first path scoring threshold LY and the second path scoring threshold LP are compared and analyzed with the path comprehensive score LJ to evaluate the operation path. The specific evaluation content is as follows:

[0040] If the path comprehensive score LJ ≤ the first path score threshold LY, it means that the operation path is qualified, and the operation continues and the operation process of the intelligent weeding robot is continuously monitored;

[0041] If the first path scoring threshold LY < path comprehensive score LJ ≤ second path scoring threshold LP, it means that there is a deviation in the operation path, and the local operation path is optimized according to the operation data set D;

[0042] If the comprehensive path score LJ> the second path score threshold LP, it means that the operation path is unqualified. The intelligent weeding robot is immediately controlled to stop the weeding operation and trigger the early warning mechanism. The early warning information is generated and sent to the intelligent weeding robot control center until a response is received from the intelligent weeding robot control center. The reason for the unqualified operation path is checked and the operation path is re-planned.

[0043] Preferably, the operation path optimization module includes an operation path planning unit and an intelligent feedback unit;

[0044] The operation path planning unit is used to perform local path planning based on the real-time operation data set D when there is a deviation in the operation path, that is, LY<LJ≤LP, using the dynamic path planning algorithm RRT-Connect, and according to the environmental data and position data in the operation data set D of the intelligent weeding robot, to obtain an operation path from the current position of the robot to the end point of the local operation area, and obtain the path adjustment amount ΔP, and the path adjustment amount ΔP is obtained in the following manner:

[0045] ΔP=P a -P c ;

[0046] Where P a represents the new operation path of the intelligent weeding robot after the operation path is optimized, P c The original operation path representing the current position of the intelligent weeding robot.

[0047] Preferably, the intelligent feedback unit is used to transmit the acquired path adjustment amount ΔP to the control center of the intelligent weeding robot, and generate an intelligent weeding robot control instruction based on the path adjustment amount ΔP, send the intelligent weeding robot control instruction to the intelligent weeding robot to adjust the operation path, and during the weeding operation, according to the obstacle data of the operation area monitored in real time by the intelligent weeding robot, when a new obstacle appears in the operation area, the operation path is adjusted in real time.

[0048] The present invention provides an all-terrain photovoltaic power station intelligent weeding robot control system based on real-time monitoring of occlusion, which has the following beneficial effects:

[0049] (1) The system can dynamically evaluate and optimize the operation path by real-time monitoring of obstacles, weeds and deviations in the operation path. In particular, by deploying lidar, ultrasonic sensors and high-definition cameras to collect operation data in real time, the system can adjust the operation path in a timely manner based on real-time environmental data to ensure that the robot can accurately avoid obstacles and adjust the operation path to avoid the impact of path deviation on operation efficiency and quality. This enables the system to improve the real-time and accuracy of weeding operations, avoid repeated operations and unnecessary downtime caused by path errors, and improve overall operation efficiency.

[0050] (2) By introducing three-dimensional modeling technology and path planning algorithms, this system can automatically plan a suitable weeding path according to the actual terrain and obstacle distribution of the photovoltaic power station. Especially in the complex photovoltaic power station environment, the intelligent weeding robot can flexibly optimize the operation path dynamically based on the real-time data collected by the sensor, such as weed coverage, the location of photovoltaic panels and the location of obstacles, and adapt to the changes in different environments in the photovoltaic power station. The dynamic path planning algorithm RRT-Connect is used to adjust the local path. The system can quickly respond to path deviations to ensure that weeding operations can be carried out smoothly in complex terrain and reduce operation interruptions caused by environmental changes. This reflects the system's adaptability and intelligence in different environments.

[0051] (3) Through a comprehensive path evaluation and optimization mechanism, this system effectively avoids possible repeated paths and invalid operations during robot operations, significantly reducing energy consumption and operating costs. During the path planning and adjustment process, the system can monitor the weed coverage, obstacle distribution and robot operation deviation in the operating area in real time, and optimize the operating path based on this, thereby reducing the movement of the robot in invalid areas and ensuring the efficiency and economy of the operation. Through reasonable path planning and dynamic adjustment, the system can optimize the use of resources, reduce unnecessary energy consumption, and improve the overall economic benefits of weeding operations. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 This is a block diagram of the control system of the all-terrain photovoltaic power station intelligent weeding robot based on real-time monitoring of occlusion in the present invention. DETAILED DESCRIPTION

[0053] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0054] Example 1

[0055] See also Figure 1 , the present invention provides an all-terrain photovoltaic power station intelligent weeding robot control system based on real-time monitoring of occlusion, including a data acquisition module, a three-dimensional model building module, a path planning module, an operation path evaluation module and an operation path optimization module;

[0056] The data acquisition module is used to install a laser radar device and a visual sensor in the photovoltaic power station, and collect point cloud data and image data of the photovoltaic power station based on the laser radar device and the visual sensor to construct an environmental data set S;

[0057] The three-dimensional model building module is used to perform preprocessing according to the environmental data set S, and to construct a three-dimensional model of the photovoltaic power station according to the preprocessed environmental data set S, and to divide the three-dimensional model of the photovoltaic power station into a plurality of grid areas;

[0058] The path planning module is used to obtain the operation path of the intelligent weeding robot based on the three-dimensional model of the photovoltaic power station using a path planning algorithm, and perform the weeding task according to the operation path;

[0059] The operation path evaluation module is used to deploy laser radar equipment, ultrasonic sensors and high-definition cameras on the intelligent weeding robot, collect the operation data of the intelligent weeding robot in real time, and construct an operation data set D. According to the operation data set D, the path comprehensive score LJ of each grid area is obtained, and the first path score threshold LY and the second path score threshold LP are preset. The first path score threshold LY and the second path score threshold LP are compared and analyzed with the path comprehensive score LJ to evaluate the operation path and obtain the grid area where the operation path has deviations;

[0060] The operation path optimization module is used to obtain the path adjustment amount ΔP according to the operation data set D for the grid area where the evaluation result shows that the operation path has deviation, and generate a control instruction according to the path adjustment amount ΔP to control the intelligent weeding robot to adjust the operation path.

[0061] In this embodiment, through the data acquisition module, the system can collect point cloud data and image data of the photovoltaic power station in real time, and build a detailed three-dimensional model of the photovoltaic power station based on the point cloud data and image data, thereby providing accurate environmental information for path planning. This accurate modeling enables the intelligent weeding robot to perform efficient and accurate weeding operations in a complex photovoltaic power station environment, avoiding the inefficiency and high risk of traditional manual weeding methods. Through the path planning module, the system can automatically plan the operation path based on the three-dimensional model of the photovoltaic power station and environmental data, and consider obstacle avoidance and operation area coverage to ensure that the intelligent weeding robot can successfully complete the weeding task. In addition, the operation path evaluation module uses the operation data collected in real time to evaluate the operation path. The robot can conduct a comprehensive path evaluation and comprehensive scoring based on the data, including distance data, lighting data, location data and weed coverage, and compare the evaluation results with the preset thresholds to timely judge the compliance of the path. This evaluation mechanism can effectively avoid deviation or failure of the operation path and improve the efficiency and accuracy of weed control. In terms of path optimization, the operation path optimization module makes real-time adjustments to the path deviation area and combines dynamic environmental data to make local path adjustments, effectively correcting the path deviation and ensuring that the robot always performs tasks along the operation path. This dynamic optimization capability enables the robot to respond to environmental changes in real time, such as new obstacles or weed growth, further improving the system's adaptability and operational flexibility.

[0062] Example 2

[0063] Please refer to Figure 1 ,Specifically: the data acquisition module includes a point cloud data acquisition unit and an image data acquisition unit;

[0064] The point cloud data acquisition unit is used to install a laser radar device in the photovoltaic power station, and use the laser radar device to monitor the photovoltaic power station in real time, obtain point cloud data of the photovoltaic power station, and construct a point cloud data group, wherein the point cloud data group includes the terrain, obstacles and point cloud data of the photovoltaic power station and photovoltaic panels;

[0065] The image data acquisition unit is used to install a visual sensor in the photovoltaic power station, and use the visual sensor to collect image data of the photovoltaic power station to construct an image data group;

[0066] The image data set includes shadow area, obstacle and weed cover image data of the photovoltaic power station;

[0067] The constructed point cloud data set and image data set are transmitted to the central database of the photovoltaic power station in real time through the wireless transmission equipment carried by the laser radar equipment and the visual sensor, and the environmental data set S is constructed.

[0068] In this embodiment, by combining laser radar and visual sensors, the advantages of both are fully utilized to ensure all-round and real-time acquisition of environmental information in the photovoltaic power station. The laser radar equipment can accurately obtain the terrain, obstacles and three-dimensional point cloud data of the photovoltaic power station and photovoltaic panels, and can efficiently identify obstacles, ups and downs on the ground and the distribution of panels. The image data acquisition unit supplements the gaps of laser radar through visual sensors, especially in the recognition of details of shadow areas, weed coverage and obstacles. Image data can provide more detailed information about the working environment, such as the type, density and growth status of weeds, to help the robot make accurate working judgments. By transmitting the point cloud data collected by the lidar and the image data collected by the visual sensor to the central database of the photovoltaic power station in real time and integrating them into the environmental data set S, the system can dynamically update and accurately reflect the real-time environmental conditions in the photovoltaic power station. The real-time transmission and storage of this data not only ensures the timeliness of data processing, but also provides high-quality original data support for the subsequent three-dimensional model building module, path planning module, operation path evaluation module and operation path optimization module. It can realize accurate information collection and environmental modeling in complex environments, and effectively improve the adaptability and reaction speed of the intelligent weeding robot to obstacles, weeds and changes in internal conditions of the photovoltaic power station. In addition, the transmission and update of real-time data ensures that various emergencies in the operation of the photovoltaic power station can be responded to quickly, thereby improving the efficiency and intelligence level of weeding operations.

[0069] Example 3

[0070] Please refer to Figure 1 ,Specifically: the three-dimensional model building module includes a data preprocessing unit and a three-dimensional model building unit;

[0071] The data preprocessing unit is used to perform point cloud data preprocessing and image data preprocessing according to the constructed environment data set S;

[0072] The point cloud data preprocessing refers to removing isolated points in the point cloud data group in the environmental data set S using a statistical outlier removal method, simplifying the point cloud data using a voxel grid method, and performing point cloud region segmentation using a region growing algorithm;

[0073] The image data preprocessing is used to remove image noise using Gaussian filtering, enhance shadow area detail information using a contrast stretching algorithm, and segment vegetation coverage area images and obstacle images using a deep learning algorithm for the image data group in the environmental data set S;

[0074] The three-dimensional model building unit is used to build a three-dimensional model of the photovoltaic power station using 3D modeling software based on the pre-processed environmental data set S, and divide the three-dimensional model of the photovoltaic power station into a plurality of grid areas, each grid area represents a weeding task point;

[0075] The photovoltaic power station 3D model is based on the point cloud data set of the photovoltaic power station, and the photovoltaic power station 3D model framework is constructed. Feature data is extracted based on the image data set of the photovoltaic power station, and the extracted feature data is mapped to the photovoltaic power station 3D model framework to obtain a complete photovoltaic power station 3D model.

[0076] The three-dimensional model of the photovoltaic power station includes the layout of photovoltaic panels, the location and height of obstacles, and the terrain changes of the photovoltaic power station.

[0077] In this embodiment, the point cloud data and image data in the environmental data set S are preprocessed to ensure that the data quality meets the requirements of subsequent modeling and path planning. The point cloud data preprocessing removes isolated points through the statistical outlier removal method to avoid erroneous modeling caused by data noise. At the same time, the voxel grid method is used to simplify the point cloud data, which reduces the complexity of processing data and makes the photovoltaic power station three-dimensional model construction process more efficient. The point cloud region segmentation using the regional growing algorithm helps to identify different areas within the photovoltaic power station, such as the distribution of obstacles and photovoltaic panels, and improves the accuracy and detail of the photovoltaic power station three-dimensional model. The image data preprocessing removes noise in the image through Gaussian filtering, effectively improves the quality of the image data, and enhances the reliability of subsequent image processing and analysis. The contrast stretching algorithm enhances the detail information of the shadow area. , ensuring that clear vegetation and obstacle information can still be obtained in weak light or complex environments. The deep learning algorithm uses intelligent image segmentation technology to accurately separate vegetation coverage areas and obstacle areas, thereby providing more accurate data support for path planning and operation path optimization. The pre-processed environmental data set S is converted into a three-dimensional model of the photovoltaic power station through the three-dimensional model construction unit, which can truly reflect the actual terrain, obstacle locations and layout of photovoltaic panels in the power station, and provide an accurate operation map for the intelligent weeding robot. On this basis, the path planning module can effectively plan the weeding path, avoid obstacles, and optimize the operation route. At the same time, the construction of the three-dimensional model of the photovoltaic power station can provide a highly accurate operation reference for the intelligent weeding robot, reducing the time and energy waste caused by improper path selection.

[0078] Example 4

[0079] Please refer to Figure 1Specifically: the path planning module is used to perform a grid search in the three-dimensional model of the photovoltaic power station using a path planning algorithm according to the three-dimensional model of the photovoltaic power station, and to perform obstacle avoidance operation path planning according to the obstacle positions in the three-dimensional model of the photovoltaic power station, and to obtain an operation path from the starting position of the intelligent weeding robot to the end point of the weeding operation task, wherein the operation path includes all weeding operation task points of the intelligent weeding robot and the coordinates of the weeding operation task points (x pre ,y pre ), and sends the operation path to the intelligent weeding robot to perform the weeding task.

[0080] In this embodiment, by using the three-dimensional model of the photovoltaic power station, on the basis of grid search, according to the actual obstacle positions and terrain features in the photovoltaic power station, an operation path from the starting position of the intelligent weeding robot to the end point of the operation task is intelligently planned. In the process of path planning, the system takes into account various possible obstacles in the photovoltaic power station, such as equipment, facilities and terrain undulations, which can effectively avoid the collision between the intelligent weeding robot and obstacles, thereby ensuring the safety of the operation. In the process of planning the operation path, the system adopts a grid search method, and calculates the position and influence of obstacles in real time based on the three-dimensional model of the photovoltaic power station, accurately identifies which areas are passable and which areas need to be avoided. This process makes path planning more accurate and can dynamically select the appropriate path according to the specific operation task requirements. The path planning module can integrate all weeding task points into the planned path to ensure that the tasks are completed one by one in sequence during the execution process, avoid duplication or omission of operations, and improve the comprehensiveness and efficiency of weeding operations. Finally, the planned path is sent to the intelligent weeding robot in real time to ensure that the robot can perform weeding tasks according to the operation path. This path planning not only improves the operation efficiency, but also ensures the robot's obstacle avoidance ability, enabling it to successfully complete the weeding task in a complex photovoltaic power station environment, reduce manual intervention, and reduce operation risks. At the same time, by avoiding unnecessary detours and ineffective operations, the system further reduces energy consumption and operating costs, and realizes efficient use of resources.

[0081] Example 5

[0082] Please refer to Figure 1 ,Specifically: the operation path evaluation module includes an operation data acquisition unit, a data analysis unit and an evaluation unit;

[0083] The operation data collection unit is used to deploy laser radar equipment, ultrasonic sensors, high-definition cameras and high-precision GPS on the intelligent weeding robot. During the weeding operation, the horizontal distance D between the intelligent weeding robot and the obstacle, the real-time weeding task point (xact ,y act ), weed height H, weed density ρ, weed area R, lateral projection area S of the obstacle and height H of the obstacle, and install a light sensor on the photovoltaic panel to collect the angle θ between the light direction and the obstacle, and based on the collected horizontal distance D between the intelligent weeding robot and the obstacle, the real-time weeding task point (x act ,y act ), weed height H, weed density ρ, weed area R, lateral projection area S of the obstacle, obstacle height H and angle θ between the light direction and the obstacle, to construct the operation data set D;

[0084] The horizontal distance D between the intelligent weeding robot and the obstacle is obtained by an ultrasonic sensor;

[0085] During the operation, the real-time weeding task point (x act ,y act ) obtained through high-precision GPS;

[0086] The height H, weed density ρ and weed area R of the weeds are obtained through a weed coverage image collected by a high-definition camera;

[0087] The lateral projection area S of the obstacle is obtained through a weed-covered image collected by a high-definition camera;

[0088] The height H of the obstacle is obtained by a laser radar device;

[0089] The angle θ between the illumination direction and the obstacle is obtained by a light sensor;

[0090] The data analysis unit is used to perform preprocessing according to the operation data set D, including denoising, outlier processing and data standardization, and perform summary calculation according to the preprocessed operation data set D to obtain the obstacle shielding coefficient ZD, weed influence coefficient ZC and path deviation coefficient PX of each grid area;

[0091] The obstacle shielding coefficient ZD is obtained in the following way:

[0092]

[0093] In the formula, H i represents the height of the ith obstacle, D i represents the horizontal distance from the position of the intelligent weeding robot to the i-th obstacle, S i represents the lateral projection area of ​​the i-th obstacle, θ irepresents the angle between the illumination direction and the i-th obstacle, n represents the total number of obstacles in the grid area, i = 1, 2, 3, ..., n;

[0094] The weed influence coefficient ZC is obtained as follows:

[0095]

[0096] In the formula, H g represents the average height of weeds in the grid area, ρ g represents the weed density in the grid area, R w is the weed area in the grid area, α and β represent the average height of weeds in the grid area, H g and the weed density ρ within the grid area g The weight coefficients are 0<α<1, 0<β<1, and α+β=1, wherein the specific values ​​of the weight coefficients α and β are set by the user according to usage.

[0097] The path deviation coefficient PX is obtained as follows:

[0098]

[0099] In the formula, x act (t) represents the x-axis coordinate of the intelligent weeding robot at the operation time point t, y act (t) represents the y-axis coordinate of the intelligent weeding robot at the operation time point t, x pre (t) represents the x-axis coordinate of the preset path of the intelligent weeding robot at the operation time point t, y pre (t) represents the y-axis coordinate of the preset path of the intelligent weeding robot at the operation time point t, represents the real-time weeding task point of the intelligent weeding robot (x act ,y act ) and the preset weeding task point (x pre ,y pre ), T represents the total operation time of the intelligent weeding robot in the grid area; t represents the t-th operation time point of the intelligent weeding robot in the grid area, t = 1, 2, 3, ..., T; wherein the preset operation point (x pre ,y pre ) is obtained through the planned operation path.

[0100] The evaluation unit is used to perform summary calculation based on the obstacle shielding coefficient ZD, the weed influence coefficient ZC and the path deviation coefficient PX to obtain the path comprehensive score LJ. The path comprehensive score LJ is obtained in the following manner:

[0101]

[0102] Wherein, δ1, δ2 and δ3 represent the weight coefficients of the obstacle shielding coefficient ZD, the weed influence coefficient ZC and the path deviation coefficient PX respectively, 0<δ1<1, 0<δ2<1, 0<δ3<1, and δ1+δ2+δ3=1, C represents the correction constant, wherein the specific values ​​of the weight coefficients δ1, δ2 and δ3 are set by the user according to the usage.

[0103] The first path scoring threshold LY and the second path scoring threshold LP are preset, and the first path scoring threshold LY and the second path scoring threshold LP are compared and analyzed with the path comprehensive score LJ to evaluate the operation path. The specific evaluation contents are as follows:

[0104] If the path comprehensive score LJ ≤ the first path score threshold LY, it means that the operation path is qualified, and the operation continues and the operation process of the intelligent weeding robot is continuously monitored;

[0105] If the first path scoring threshold LY < path comprehensive score LJ ≤ second path scoring threshold LP, it means that there is a deviation in the operation path, and the local operation path is optimized according to the operation data set D;

[0106] If the comprehensive path score LJ> the second path score threshold LP, it means that the operation path is unqualified. The intelligent weeding robot is immediately controlled to stop the weeding operation and trigger the early warning mechanism. The early warning information is generated and sent to the intelligent weeding robot control center until a response is received from the intelligent weeding robot control center. The reason for the unqualified operation path is checked and the operation path is re-planned.

[0107] In this embodiment, by integrating multiple sensors and data analysis technologies, the robot's operating path can be monitored and optimized in real time to improve operating efficiency and safety. First, the operating data is collected in real time through laser radar, ultrasonic sensor, high-definition camera and high-precision GPS, covering obstacle distance, weed coverage and light data. These data are continuously updated during the operation process and used to calculate the obstacle occlusion coefficient ZD, weed influence coefficient ZC and path deviation coefficient PX, providing a scientific basis for path evaluation. By integrating these data, the system can accurately judge the compliance of the current path, adjust the operating strategy in time, and ensure that the robot continues to work efficiently in a complex environment. Secondly, the system can dynamically handle changes in the operating environment, such as obstacle occlusion and weed growth, avoiding the static limitations of traditional path planning methods. The real-time calculation of the obstacle occlusion coefficient ZD and the weed influence coefficient ZC enables the robot to automatically optimize the path and ensure the operating quality when facing sudden obstacles or weed-dense areas. This adaptability not only improves the weeding effect, It also reduces the operation stagnation caused by path deviation or environmental changes. In addition, the real-time monitoring and optimization mechanism of the path deviation coefficient PX ensures that the robot always works along the operation path, avoiding resource waste or equipment damage caused by path errors. Especially in complex scenarios such as photovoltaic power stations, the system can intelligently adjust the path and improve operation efficiency by monitoring the lighting conditions through light sensors and combining them with other operation data. The system also has a multi-level path evaluation and early warning mechanism, which can evaluate the operation path in real time according to the comprehensive path score LJ. If the comprehensive path score LJ is unqualified, the system will ensure operation safety through intelligent optimization or shutdown early warning. This intelligent path management mechanism not only reduces the need for manual intervention, but also improves the robot's operation accuracy and safety, ensuring the continuity and efficiency of the operation. In short, through intelligent path evaluation and optimization, the system significantly improves the operation efficiency, accuracy and adaptability of the weeding robot, provides strong technical support for weeding operations in complex environments, reduces manual intervention, and improves work safety.

[0108] Example 6

[0109] Please refer to Figure 1 ,Specifically: the operation path optimization module includes an operation path planning unit and an intelligent feedback unit;

[0110] The operation path planning unit is used to perform local path planning based on the real-time operation data set D when there is a deviation in the operation path, that is, LY<LJ≤LP, using the dynamic path planning algorithm RRT-Connect, and according to the environmental data and position data in the operation data set D of the intelligent weeding robot, to obtain an operation path from the current position of the robot to the end point of the local operation area, and obtain the path adjustment amount ΔP, and the path adjustment amount ΔP is obtained in the following manner:

[0111] ΔP=P a -P c ;

[0112] Where P a represents the new operation path of the intelligent weeding robot after the operation path is optimized, P c The original operation path representing the current position of the intelligent weeding robot.

[0113] The intelligent feedback unit is used to transmit the acquired path adjustment amount ΔP to the control center of the intelligent weeding robot, and generate an intelligent weeding robot control instruction based on the path adjustment amount ΔP, and send the intelligent weeding robot control instruction to the intelligent weeding robot to adjust the operation path, and during the weeding operation, according to the obstacle data of the operation area monitored in real time by the intelligent weeding robot, when a new obstacle appears in the operation area, the operation path is adjusted in real time.

[0114] In this embodiment, by introducing the dynamic path planning algorithm RRT-Connect, the operation path planning unit can quickly perform local path planning based on the real-time acquired environmental data and position data when the operation path of the intelligent weeding robot deviates. The algorithm has strong flexibility and adaptability, and can automatically adjust the path in a complex and dynamically changing operation environment to avoid operation interruption or unnecessary repeated operations caused by path deviation, thereby significantly improving the operation efficiency and path accuracy. Secondly, the intelligent feedback unit can transmit the path adjustment information to the control center in real time, generate corresponding control instructions, and immediately feedback to the robot to ensure that the path adjustment instructions are executed in the shortest time. This high degree of real-time performance and rapid response ensures that the robot can quickly respond to path deviations and environmental changes, and will not cause a decrease in operation efficiency or robot operation errors due to delayed adjustment, thereby improving the accuracy and flexibility of the operation path of the intelligent weeding robot. The real-time path adjustment and feedback mechanism also improves the robot's adaptability to dynamic environments, enabling it to efficiently perform operation tasks under complex and changeable operating conditions. These advantages combined make the intelligent weeding robot more efficient and energy-saving in weeding operations in photovoltaic power stations, and reduce operating costs, further promoting the integration and development of agriculture and automation technology.

[0115] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An all-terrain photovoltaic power station intelligent weeding robot control system based on real-time monitoring of occlusion, characterized by: It includes data acquisition module, 3D model building module, path planning module, operation path evaluation module and operation path optimization module; The data acquisition module is used to install a laser radar device and a visual sensor in the photovoltaic power station, and collect point cloud data and image data of the photovoltaic power station based on the laser radar device and the visual sensor to construct an environmental data set S; The three-dimensional model building module is used to perform preprocessing according to the environmental data set S, and to construct a three-dimensional model of the photovoltaic power station according to the preprocessed environmental data set S, and to divide the three-dimensional model of the photovoltaic power station into a plurality of grid areas; The path planning module is used to obtain the operation path of the intelligent weeding robot based on the three-dimensional model of the photovoltaic power station using a path planning algorithm, and perform the weeding task according to the operation path; The operation path evaluation module is used to deploy laser radar equipment, ultrasonic sensors and high-definition cameras on the intelligent weeding robot, collect the operation data of the intelligent weeding robot in real time, and construct an operation data set D. According to the operation data set D, the path comprehensive score LJ of each grid area is obtained, and the first path score threshold LY and the second path score threshold LP are preset. The first path score threshold LY and the second path score threshold LP are compared and analyzed with the path comprehensive score LJ to evaluate the operation path and obtain the grid area where the operation path has deviations; The operation path optimization module is used to obtain the path adjustment amount ΔP according to the operation data set D for the grid area where the evaluation result shows that the operation path has deviation, and generate a control instruction according to the path adjustment amount ΔP to control the intelligent weeding robot to adjust the operation path.

2. The all-terrain photovoltaic power station intelligent weeding robot control system based on real-time monitoring of occlusion according to claim 1 is characterized by: The data acquisition module includes a point cloud data acquisition unit and an image data acquisition unit; The point cloud data acquisition unit is used to install a laser radar device in the photovoltaic power station, and use the laser radar device to monitor the photovoltaic power station in real time, obtain point cloud data of the photovoltaic power station, and construct a point cloud data group, wherein the point cloud data group includes the terrain, obstacles and point cloud data of the photovoltaic power station and photovoltaic panels; The image data acquisition unit is used to install a visual sensor in the photovoltaic power station, and use the visual sensor to collect image data of the photovoltaic power station to construct an image data group; The image data set includes shadow area, obstacles and weed cover image data of the photovoltaic power station; The constructed point cloud data set and image data set are transmitted to the central database of the photovoltaic power station in real time through the wireless transmission equipment carried by the laser radar equipment and the visual sensor, and the environmental data set S is constructed.

3. The all-terrain photovoltaic power station intelligent weeding robot control system based on real-time monitoring of occlusion according to claim 2 is characterized by: The three-dimensional model building module includes a data preprocessing unit and a three-dimensional model building unit; The data preprocessing unit is used to perform point cloud data preprocessing and image data preprocessing according to the constructed environment data set S; The point cloud data preprocessing refers to removing isolated points in the point cloud data group in the environmental data set S using a statistical outlier removal method, simplifying the point cloud data using a voxel grid method, and performing point cloud region segmentation using a region growing algorithm; The image data preprocessing is used to remove image noise using Gaussian filtering, enhance shadow area detail information using a contrast stretching algorithm, and segment vegetation coverage area images and obstacle images using a deep learning algorithm for the image data group in the environmental data set S; The three-dimensional model building unit is used to build a three-dimensional model of the photovoltaic power station using 3D modeling software based on the pre-processed environmental data set S, and divide the three-dimensional model of the photovoltaic power station into a plurality of grid areas, each grid area represents a weeding task point; The photovoltaic power station 3D model is based on the point cloud data set of the photovoltaic power station, and the photovoltaic power station 3D model framework is constructed. Feature data is extracted based on the image data set of the photovoltaic power station, and the extracted feature data is mapped to the photovoltaic power station 3D model framework to obtain a complete photovoltaic power station 3D model. The three-dimensional model of the photovoltaic power station includes the layout of photovoltaic panels, the location and height of obstacles, and the terrain changes of the photovoltaic power station.

4. The all-terrain photovoltaic power station intelligent weeding robot control system based on real-time occlusion monitoring according to claim 3 is characterized by: The path planning module is used to perform a grid search in the three-dimensional model of the photovoltaic power station using a path planning algorithm according to the three-dimensional model of the photovoltaic power station, and to perform obstacle avoidance operation path planning according to the obstacle positions in the three-dimensional model of the photovoltaic power station, and to obtain an operation path from the starting position of the intelligent weeding robot to the end point of the weeding operation task, wherein the operation path includes all weeding operation task points of the intelligent weeding robot and the coordinates (x pre ,y pre ), and sends the operation path to the intelligent weeding robot to perform the weeding task.

5. The all-terrain photovoltaic power station intelligent weeding robot control system based on real-time occlusion monitoring according to claim 1 is characterized in that: The operation path evaluation module includes an operation data acquisition unit, a data analysis unit and an evaluation unit; The operation data collection unit is used to deploy laser radar equipment, ultrasonic sensors, high-definition cameras and high-precision GPS on the intelligent weeding robot. During the weeding operation, the horizontal distance D between the intelligent weeding robot and the obstacle, the real-time weeding task point (x act ,y act ), weed height H, weed density ρ, weed area R, lateral projection area S of the obstacle and height H of the obstacle, and install a light sensor on the photovoltaic panel to collect the angle θ between the light direction and the obstacle, and based on the collected horizontal distance D between the intelligent weeding robot and the obstacle, the real-time weeding task point (x act ,y act ), weed height H, weed density ρ, weed area R, lateral projection area S of the obstacle, obstacle height H and angle θ between the light direction and the obstacle, to construct the operation data set D; The data analysis unit is used to perform preprocessing according to the operation data set D, including denoising, outlier processing and data standardization, and perform summary calculation according to the preprocessed operation data set D to obtain the obstacle shielding coefficient ZD, weed influence coefficient ZC and path deviation coefficient PX of each grid area; The obstacle shielding coefficient ZD is obtained in the following way: In the formula, H i represents the height of the ith obstacle, D i represents the horizontal distance from the position of the intelligent weeding robot to the i-th obstacle, S i represents the lateral projection area of ​​the i-th obstacle, θ i represents the angle between the illumination direction and the i-th obstacle, n represents the total number of obstacles in the grid area, i = 1, 2, 3, ..., n; The weed influence coefficient ZC is obtained as follows: In the formula, H g represents the average height of weeds in the grid area, ρ g represents the weed density in the grid area, R w is the weed area in the grid area, α and β represent the average height of weeds in the grid area, H g and the weed density ρ within the grid area g The weight coefficient is 0<α<1, 0<β<1, and α+β=1.

6. The all-terrain photovoltaic power station intelligent weeding robot control system based on real-time occlusion monitoring according to claim 5 is characterized by: The path deviation coefficient PX is obtained as follows: In the formula, x act (t) represents the x-axis coordinate of the intelligent weeding robot at the operation time point t, y act (t) represents the y-axis coordinate of the intelligent weeding robot at the operation time point t, x pre (t) represents the x-axis coordinate of the preset path of the intelligent weeding robot at the operation time point t, y pre (t) represents the y-axis coordinate of the preset path of the intelligent weeding robot at the operation time point t, represents the real-time weeding task point of the intelligent weeding robot (x act ,y act ) and the preset weeding task point (x pre ,y pre ), T represents the total operation time of the intelligent weeding robot in the grid area; t represents the t-th operation time point of the intelligent weeding robot in the grid area, t = 1, 2, 3, ..., T; wherein the preset operation point (x pre ,y pre ) is obtained through the planned operation path.

7. The all-terrain photovoltaic power station intelligent weeding robot control system based on real-time occlusion monitoring according to claim 6 is characterized by: The evaluation unit is used to perform summary calculation based on the obstacle shielding coefficient ZD, the weed influence coefficient ZC and the path deviation coefficient PX to obtain the path comprehensive score LJ. The path comprehensive score LJ is obtained in the following manner: Wherein, δ1, δ2 and δ3 represent the weight coefficients of the obstacle shielding coefficient ZD, the weed influence coefficient ZC and the path deviation coefficient PX, respectively, 0<δ1<1, 0<δ2<1, 0<δ3<1, and δ1+δ2+δ3=1, C represents the correction constant.

8. The all-terrain photovoltaic power station intelligent weeding robot control system based on real-time occlusion monitoring according to claim 7 is characterized by: The first path scoring threshold LY and the second path scoring threshold LP are preset, and the first path scoring threshold LY and the second path scoring threshold LP are compared and analyzed with the path comprehensive score LJ to evaluate the operation path. The specific evaluation contents are as follows: If the path comprehensive score LJ ≤ the first path score threshold LY, it means that the operation path is qualified, and the operation continues and the operation process of the intelligent weeding robot is continuously monitored; If the first path scoring threshold LY < path comprehensive score LJ ≤ second path scoring threshold LP, it means that there is a deviation in the operation path, and the local operation path is optimized according to the operation data set D; If the comprehensive path score LJ> the second path score threshold LP, it means that the operation path is unqualified. The intelligent weeding robot is immediately controlled to stop the weeding operation and trigger the early warning mechanism. The early warning information is generated and sent to the intelligent weeding robot control center until a response is received from the intelligent weeding robot control center. The reason for the unqualified operation path is checked and the operation path is re-planned.

9. The all-terrain photovoltaic power station intelligent weeding robot control system based on real-time occlusion monitoring according to claim 8, characterized in that: The operation path optimization module includes an operation path planning unit and an intelligent feedback unit; The operation path planning unit is used to perform local path planning based on the real-time operation data set D when there is a deviation in the operation path, that is, LY<LJ≤LP, using the dynamic path planning algorithm RRT-Connect, and according to the environmental data and position data in the operation data set D of the intelligent weeding robot, to obtain an operation path from the current position of the robot to the end point of the local operation area, and obtain the path adjustment amount ΔP, and the path adjustment amount ΔP is obtained in the following manner: ΔP=P a -P c ; Where P a represents the new operation path of the intelligent weeding robot after the operation path is optimized, P c The original operation path representing the current position of the intelligent weeding robot.

10. The all-terrain photovoltaic power station intelligent weeding robot control system based on real-time occlusion monitoring according to claim 9, characterized in that: The intelligent feedback unit is used to transmit the acquired path adjustment amount ΔP to the control center of the intelligent weeding robot, and generate an intelligent weeding robot control instruction based on the path adjustment amount ΔP, and send the intelligent weeding robot control instruction to the intelligent weeding robot to adjust the operation path, and during the weeding operation, according to the obstacle data of the operation area monitored in real time by the intelligent weeding robot, when a new obstacle appears in the operation area, the operation path is adjusted in real time.

Images