A multi-robot collaborative photovoltaic panel cleaning control method and system

Abstract

The present invention provides a multi-robot collaborative photovoltaic panel cleaning control method and system, which relates to the technical field of automated cleaning robots. The method includes receiving a cleaning task request from a monitoring end, retrieving a virtual photovoltaic space of an actual photovoltaic station; collecting a surface image of the actual photovoltaic panel and inputting it into a snow detection model, identifying the actual snow area and determining characteristic parameters, and obtaining meteorological data through meteorological monitoring equipment; determining whether to trigger an extreme snow cleaning mode based on the characteristic parameters and meteorological data; if triggered, dividing the actual photovoltaic station into multiple cleaning sub-areas and assigning them to multiple robots, determining snow cleaning paths and parameters; the robots perform tasks according to the snow cleaning paths and parameters, and monitor the surface cleanliness of the actual photovoltaic panels; adjusting the snow cleaning parameters based on the surface cleanliness and updating the virtual photovoltaic space, thereby improving the efficiency of multi-robot collaborative cleaning of photovoltaic panels under extreme weather conditions and achieving precise adjustment and optimization of the cleaning mode.

Description

Technical Field

[0001] The present invention relates to the technical field of automated cleaning robots, and in particular to a multi-robot collaborative photovoltaic panel cleaning control method and system. Background Art

[0002] In the field of automated cleaning, especially in the photovoltaic panel cleaning industry, multi-robot collaborative operation has become the current development trend to improve cleaning efficiency and reduce manual intervention.

[0003] While existing multi-robot cleaning control methods can effectively improve cleaning efficiency in conventional environments, they typically rely on pre-set path planning and simple task allocation mechanisms to achieve multi-robot coordination in specific scenarios, such as photovoltaic panel areas. In extreme weather conditions, such as when panels are covered in heavy snow, existing cleaning methods struggle to effectively cope. This not only significantly reduces panel cleaning efficiency but can even render the cleaning task impossible due to excessive snow accumulation, seriously impacting the panels' power generation efficiency.

[0004] Therefore, it is necessary to provide a multi-robot collaborative photovoltaic panel cleaning control method and system to solve the above technical problems. Summary of the Invention

[0005] In order to solve the above technical problems, the present invention provides a multi-robot collaborative photovoltaic panel cleaning control method and system, which is used to solve the problem that the existing photovoltaic panel cleaning control method has low efficiency of multi-robot collaborative cleaning under extreme heavy snow conditions and is difficult to effectively adjust and optimize the cleaning mode.

[0006] The present invention provides a multi-robot collaborative photovoltaic panel cleaning control method, the method comprising:

[0007] Receive cleaning task requests from the monitoring end and retrieve the virtual photovoltaic space corresponding to the actual photovoltaic site;

[0008] Collecting photovoltaic surface images of actual photovoltaic panels in the actual photovoltaic station and inputting them into a snow detection model, identifying actual snow areas and determining corresponding snow characteristic parameters through the snow detection model, and simultaneously obtaining real-time meteorological data through meteorological monitoring equipment;

[0009] determining whether to trigger an extreme snow clearing mode of the robot based on the snow characteristic parameters and the real-time meteorological data;

[0010] If the extreme snow clearing mode of the robot is triggered, the actual photovoltaic station is divided into a plurality of clearing sub-areas based on the virtual photovoltaic space and the sub-areas are assigned to a plurality of robots, and the snow clearing paths and snow clearing parameters of the robots are determined at the same time;

[0011] The robot performs the snow clearing task according to the snow clearing path and the snow clearing parameters, while monitoring the surface cleanliness of the actual photovoltaic panel;

[0012] Based on the surface cleanliness, the snow removal parameters are adjusted and the virtual photovoltaic space is updated.

[0013] Preferably, the virtual photovoltaic space includes virtual photovoltaic panels corresponding to the actual photovoltaic panels, and the coordinate system of the virtual photovoltaic panels is the same as that of the actual photovoltaic panels.

[0014] Preferably, identifying the actual snow area and determining the corresponding snow characteristic parameters by the snow detection model specifically includes:

[0015] Setting a snow pixel value interval, and extracting pixel points in the photovoltaic surface image whose pixel values ​​are within the snow pixel value interval as snow pixel points through the snow detection model;

[0016] Clustering the adjacent snow pixels into a snow pixel set, and determining a snow boundary contour of the actual snow area based on the snow pixel set;

[0017] Calculating the snow accumulation area, snow accumulation area perimeter, and snow distribution range of the actual snow accumulation area based on the snow accumulation boundary contour as basic snow accumulation characteristic parameters;

[0018] Collecting snow distance data of the actual snow area through a lidar sensor, and calculating the average snow thickness and snow thickness distribution gradient in the actual snow area based on the snow distance data as core snow characteristic parameters;

[0019] Summarizing the basic snow characteristic parameters and the core snow characteristic parameters to form the snow characteristic parameters;

[0020] The snow thickness distribution gradient is used to reflect the changing trend of snow thickness in the actual snow area.

[0021] Preferably, the determining whether to trigger the robot's extreme snow clearing mode based on the snow characteristic parameters and the real-time meteorological data specifically includes:

[0022] Setting the judgment conditions for the extreme snow clearing mode, including a critical value for snow thickness, a critical value for snowfall, a critical value for wind speed, and a critical value for low temperature duration;

[0023] Comparing the average snow thickness in the snow characteristic parameter with the snow thickness critical value, and comparing the hourly snowfall, real-time wind speed, and real-time low temperature duration in the real-time meteorological data with the snowfall critical value, the wind speed critical value, and the low temperature duration critical value, respectively;

[0024] If the average snow thickness exceeds the snow thickness threshold, the hourly snowfall exceeds the snowfall threshold, the real-time wind speed exceeds the preset wind speed threshold, and the real-time low temperature duration exceeds the low temperature duration threshold, then the extreme snow clearing mode of the robot is triggered; otherwise, the normal snow clearing mode of the robot is triggered.

[0025] The snow clearing mode determination result is recorded and associated with the corresponding snow clearing strategy library, and the snow clearing strategy library includes the working parameter configuration and cooperation rules of the robot under the corresponding snow clearing mode.

[0026] Preferably, the actual photovoltaic station is divided into a plurality of cleaning sub-areas based on the virtual photovoltaic space and the cleaning sub-areas are allocated to a plurality of robots, specifically comprising:

[0027] The virtual photovoltaic space is marked with arrangement information of an actual photovoltaic panel array;

[0028] Dividing the virtual photovoltaic space into grids according to the arrangement direction of the actual photovoltaic panel array to form initial grid units;

[0029] Marking the initial grid cells covered with snow as cells to be cleaned according to the projection position of the actual snow area in the virtual photovoltaic space;

[0030] According to the preset division direction, that is, the direction from the upper edge to the lower edge of the actual photovoltaic panel, the virtual photovoltaic space is traversed, the total number of pixels in each row and the fraction of pixels in each row belonging to the unit to be cleaned are counted, and the ratio of the fraction to the total number is calculated to obtain the ratio of cleaned pixels;

[0031] When the proportions of the cleaned pixels in a plurality of consecutive rows are all greater than or equal to a preset proportion threshold, a horizontal dividing line is generated, and the generation operation of the horizontal dividing line is repeated until the virtual photovoltaic space is divided into a plurality of longitudinally arranged cleaning sub-areas;

[0032] A safety zone is provided between adjacent cleaning sub-areas, and the width of the safety zone is greater than or equal to the width of the robot body;

[0033] The cleaning sub-areas are merged or split according to the distribution density of the actual snow area until the area difference between any two cleaning sub-areas is within a preset area difference range.

[0034] Preferably, the current status information of the robot is collected in real time, including the current remaining power, current position coordinates, current workload rate, historical cleaning success rate and current device health level;

[0035] constructing a snow-clearing capability evaluation score for the robot based on the current state information, wherein the snow-clearing capability evaluation score is used to quantitatively evaluate the current snow-clearing capability of the robot, and the snow-clearing capability evaluation score is obtained by weighting the current battery sufficiency score, the current position adaptability score, the current load bearing capacity score, and the snow-clearing efficiency stability score of the robot;

[0036] Calculating snow clearing priority scores for the cleaning sub-areas, and prioritizing the cleaning sub-areas from largest to smallest according to the snow clearing priority scores, wherein the snow clearing priority scores are weighted by the snow thickness ranking scores, the area ranking scores, the photovoltaic panel power generation capacity scores, and the distance ranking scores between the sub-areas and the robot;

[0037] According to the snow clearing capability evaluation score and the snow clearing priority score, the cleaning sub-area is assigned to the corresponding robot using a dynamic matching rule, and an allocation comparison table of the cleaning sub-areas and the robots is generated and synchronized to the virtual photovoltaic space for visual marking, wherein the marking content includes the current allocation status and the estimated cleaning completion time;

[0038] The snow clearing load rates of the robots are balanced during the allocation process until the difference between the snow clearing load rates of any two robots is within a preset balance difference range.

[0039] Preferably, the process of determining the snow clearing path of the robot is as follows:

[0040] generating an initial straight path for the robot based on a virtual starting position of the robot in the virtual photovoltaic space and a virtual arrival position corresponding to the cleaning sub-area;

[0041] calling an obstacle database pre-stored in the virtual photovoltaic space to perform obstacle detection on the initial straight path, wherein the obstacle database includes three-dimensional coordinate information of static obstacles;

[0042] If there is an obstacle in the initial straight path, a multi-segment broken line obstacle avoidance path is generated using the obstacle avoidance rule and the initial straight path is adjusted to generate the snow clearing path until the distance between the snow clearing path and the obstacle is greater than or equal to the preset safety distance.

[0043] Preferably, based on the snow thickness distribution gradient in the snow characteristic parameter, a path segment in the snow clearing path where the snow thickness distribution gradient exceeds a preset gradient threshold is marked as a key snow clearing path segment, and the key snow clearing path segment accounts for greater than or equal to 60% of the path length of the snow clearing path;

[0044] The robots share the snow-clearing paths through wireless communication. If it is detected that the snow-clearing paths of different robots overlap in the same time period, the overlapping paths of the different snow-clearing paths need to be adjusted until there is no overlapping path in the same time period.

[0045] Preferably, the updating process of the virtual photovoltaic space is as follows:

[0046] After the robot completes the snow clearing task in the corresponding cleaning sub-area, it uploads a cleaning completion signal to the control center, and simultaneously monitors the surface cleanliness of the actual photovoltaic panel and transmits the signal to the control center;

[0047] The control center marks the cleaned sub-area in the virtual photovoltaic space as cleaned according to the cleaning completion signal, and the remaining cleaning sub-areas remain in a to-be-cleaned state;

[0048] The control center associates the surface cleanliness with a virtual photovoltaic panel in the virtual photovoltaic space, and visually marks the virtual photovoltaic panel using a color gradient;

[0049] The control center synchronously updates the real-time virtual position of the robot in the virtual photovoltaic space;

[0050] Associating the start time and end time of the robot performing the snow clearing task with the time axis of the virtual photovoltaic space to form a dynamic snow clearing process record;

[0051] The virtual photovoltaic space supports remote access operations of the monitoring terminal.

[0052] A multi-robot collaborative photovoltaic panel cleaning control system, the system comprising:

[0053] The photovoltaic space retrieval module is used to receive the cleaning task request from the monitoring end and retrieve the virtual photovoltaic space corresponding to the actual photovoltaic station;

[0054] A snow area recognition module is used to collect photovoltaic surface images of actual photovoltaic panels in the actual photovoltaic station and input them into a snow detection model. The snow detection model is used to identify the actual snow area and determine the corresponding snow characteristic parameters. At the same time, real-time meteorological data is obtained through meteorological monitoring equipment.

[0055] a cleaning mode determination module, configured to determine whether to trigger the robot's extreme snow cleaning mode based on the snow characteristic parameters and the real-time meteorological data;

[0056] a cleaning area allocation module, configured to, if the extreme snow cleaning mode of the robot is triggered, divide the actual photovoltaic station into a plurality of cleaning sub-areas based on the virtual photovoltaic space and allocate the sub-areas to the plurality of robots, and simultaneously determine the snow cleaning paths and snow cleaning parameters of the robots;

[0057] a cleaning task execution module, configured for the robot to perform the snow cleaning task according to the snow cleaning path and the snow cleaning parameters, while monitoring the actual surface cleanliness of the photovoltaic panel;

[0058] A photovoltaic space update module is used to adjust the snow removal parameters and update the virtual photovoltaic space based on the surface cleanliness.

[0059] Compared with related technologies, the multi-robot collaborative photovoltaic panel cleaning control method and system provided by the present invention has the following beneficial effects:

[0060] The present invention receives a cleaning task request from a monitoring end, and retrieves a virtual photovoltaic space corresponding to an actual photovoltaic station; collects photovoltaic surface images of actual photovoltaic panels in the actual photovoltaic station and inputs them into a snow detection model, identifies the actual snow area through the snow detection model and determines the corresponding snow characteristic parameters, and simultaneously obtains real-time meteorological data through meteorological monitoring equipment; determines whether to trigger the robot's extreme snow cleaning mode based on the snow characteristic parameters and real-time meteorological data; if the robot's extreme snow cleaning mode is triggered, the actual photovoltaic station is divided into multiple cleaning sub-areas based on the virtual photovoltaic space and assigned to multiple robots, and the robots' snow cleaning paths and snow cleaning parameters are determined; the robots perform snow cleaning tasks according to the snow cleaning paths and snow cleaning parameters, and simultaneously monitor the surface cleanliness of the actual photovoltaic panels; based on the surface cleanliness, the snow cleaning parameters are adjusted and the virtual photovoltaic space is updated, thereby improving the efficiency of multiple robots in collaborative cleaning of photovoltaic panels under extreme weather conditions, and achieving precise adjustment and optimization of the cleaning mode.

[0061] The present invention achieves intelligent triggering of extreme snow-clearing mode by constructing a precise mapping between virtual photovoltaic space and actual photovoltaic stations, combining snow detection models with real-time meteorological data, solving the problem of low efficiency of multi-robot collaboration in extreme weather conditions and significantly improving the response speed of photovoltaic panel cleaning. The method of the present invention is based on a task allocation mechanism based on dynamic division of sub-areas in virtual space and capability matching, which balances the robot load. Combined with the obstacle avoidance and anti-overlap design in path planning, it reduces ineffective robot operations and improves the overall snow-clearing efficiency. The present invention dynamically adjusts the robot's snow-clearing parameters based on snow characteristic parameters, and relies on virtual space to update the snow-clearing status in real time, achieving refined control of the snow-clearing process, significantly improving the compliance rate of photovoltaic panel cleanliness, and ensuring normal power generation of photovoltaic panels. By integrating dynamic adjustment and collaborative mechanisms, the present invention enhances the adaptability of multiple robots to complex snow environments, reduces equipment losses, and reduces the cost of manual intervention. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 This is a flow chart of a multi-robot collaborative photovoltaic panel cleaning control method of the present invention;

[0063] Figure 2 This is a system block diagram of a multi-robot collaborative photovoltaic panel cleaning control system of the present invention. DETAILED DESCRIPTION

[0064] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0065] Example 1

[0066] like Figure 1 As shown, a multi-robot collaborative photovoltaic panel cleaning control method includes:

[0067] Receive cleaning task requests from the monitoring end and retrieve the virtual photovoltaic space corresponding to the actual photovoltaic site;

[0068] The monitoring terminal is a control terminal capable of data collection, status monitoring, and command issuance. It is typically deployed in the photovoltaic plant's operation and maintenance center, responsible for initiating cleaning tasks and receiving feedback. The actual photovoltaic plant is the physical power generation area consisting of a large number of photovoltaic panel arrays, brackets, cables, and supporting facilities, and serves as the physical carrier of the cleaning task.

[0069] Virtual photovoltaic space refers to a digital mapping model built based on the three-dimensional layout, equipment parameters and environmental characteristics of the actual photovoltaic site. It contains virtual photovoltaic panels that correspond one-to-one with the actual photovoltaic panels. This virtual photovoltaic space can synchronize the state changes of the physical space in real time, support path simulation, task rehearsal and other functions, and is the core digital platform for realizing remote control and collaborative decision-making.

[0070] By collecting surface images of actual photovoltaic panels and inputting them into the snow detection model for analysis, the actual snow accumulation area can be accurately identified and its characteristic parameters can be extracted. At the same time, with the help of meteorological monitoring equipment, real-time meteorological data can be obtained to form a comprehensive perception of the snow status and environmental conditions, providing data support for subsequent model judgment.

[0071] Collecting photovoltaic surface images of actual photovoltaic panels in the actual photovoltaic station and inputting them into a snow detection model, identifying actual snow areas and determining corresponding snow characteristic parameters through the snow detection model, and simultaneously obtaining real-time meteorological data through meteorological monitoring equipment;

[0072] It can be understood that the photovoltaic surface image refers to the image information of the photovoltaic panel surface collected by visual sensors, such as high-definition cameras, which specifically includes visual features such as the snow coverage area and distribution pattern. The snow detection model is an intelligent analysis model based on image recognition technology. It distinguishes between snow-covered areas and non-snow-covered areas by extracting features such as the color, texture, and reflectivity of snow in the photovoltaic surface image, and has the ability to accurately identify the boundaries of snow. The actual snow area refers to the physical area on the photovoltaic panel surface covered with snow. Its boundaries and range are extracted and determined from the photovoltaic surface image by the snow detection model. Snow characteristic parameters refer to a set of quantitative indicators used to describe the state of snow, including snow coverage area, distribution range, average thickness, thickness gradient, etc., which can reflect the physical properties and distribution characteristics of snow and provide data support for the formulation of snow removal strategies.

[0073] Meteorological monitoring equipment refers to sensor devices used to collect environmental parameters such as temperature, snowfall, wind speed, and humidity. These sensors, such as temperature and humidity sensors, anemometers, and rain and snow sensors, are deployed at key locations in photovoltaic plants. Real-time meteorological data, obtained through meteorological monitoring equipment, reflects current and short-term weather trends and is a key indicator for determining extreme weather conditions.

[0074] Based on the acquired snow characteristic parameters and real-time meteorological data, a comprehensive assessment can be performed to determine whether to trigger the extreme snow clearing mode. This judgment process aims to switch to an appropriate clearing strategy based on the severity of the environment to ensure that the robot can still maintain efficient operation under extreme conditions.

[0075] determining whether to trigger an extreme snow clearing mode of the robot based on the snow characteristic parameters and the real-time meteorological data;

[0076] The extreme snow clearing mode refers to a special clearing mode designed for heavy snow environments that exceed the conventional clearing capabilities. Compared with the conventional mode, this mode has a higher coordination frequency, a wider cleaning coverage area and stronger equipment output power, and is used to deal with complex scenarios such as thick snow and continuous snowfall.

[0077] When the extreme snow-clearing mode is triggered, the actual photovoltaic site can be divided into areas based on the virtual photovoltaic space to form multiple independent cleaning sub-areas, and these sub-areas can be reasonably allocated to multiple robots. At the same time, exclusive snow-clearing paths and cleaning parameters can be planned for each robot to achieve refined allocation of cleaning tasks and personalized customization of operation plans.

[0078] If the extreme snow clearing mode of the robot is triggered, the actual photovoltaic station is divided into a plurality of clearing sub-areas based on the virtual photovoltaic space and the sub-areas are assigned to a plurality of robots, and the snow clearing paths and snow clearing parameters of the robots are determined at the same time;

[0079] It should be noted that the cleaning sub-area refers to several independent operation units formed by dividing the actual photovoltaic station based on the virtual photovoltaic space. Each unit contains a complete photovoltaic panel group, and the boundaries are set with safety intervals to avoid collisions during robot operation.

[0080] The snow-clearing path is the robot's trajectory from its starting position to the target sub-area. It is generated through virtual PV space planning and must avoid obstacles and minimize unnecessary round trips. Snow-clearing parameters are a set of operational parameters that influence clearing performance, including clearing speed, clearing width, clearing force, and operation frequency. These parameters are dynamically adjusted based on snow thickness, hardness, and PV panel material.

[0081] The robot performs the snow clearing task according to the snow clearing path and the snow clearing parameters, while monitoring the surface cleanliness of the actual photovoltaic panel;

[0082] Among them, surface cleanliness refers to an indicator that reflects the degree of snow removal on the surface of photovoltaic panels. It is evaluated by visual sensors or transmittance detection devices. It is the core standard for measuring the cleaning effect and directly affects the degree of recovery of the photovoltaic panel's power generation efficiency.

[0083] Based on the surface cleanliness, the snow removal parameters are adjusted and the virtual photovoltaic space is updated.

[0084] Finally, based on the surface cleanliness monitoring results, the cleaning parameters can be dynamically adjusted to optimize the cleaning effect. At the same time, the real-time operation status and cleanliness data can be synchronously updated to the virtual photovoltaic space to achieve accurate mapping of virtual and reality, forming a complete closed loop from task initiation, environmental perception, decision-making and planning, execution feedback to dynamic optimization, ensuring the efficiency and adaptability of multi-robot collaborative cleaning.

[0085] In a specific implementation process, the virtual photovoltaic space includes virtual photovoltaic panels corresponding to the actual photovoltaic panels, and the coordinate system of the virtual photovoltaic panels is the same as that of the actual photovoltaic panels.

[0086] A virtual photovoltaic panel refers to a digital unit in the virtual photovoltaic space that corresponds one-to-one to an actual photovoltaic panel. It carries information such as the coordinates, material, and status of the actual photovoltaic panel.

[0087] It's important to note that the virtual photovoltaic space contains virtual panels that correspond one-to-one with actual panels, maintaining the same coordinate system. As digital units within the virtual space, virtual panels carry information such as the coordinates, material, and status of the actual panels, enabling precise digital mapping of the panels.

[0088] This correspondence and coordinate consistency ensures that various operation plans for photovoltaic panels in the virtual photovoltaic space can be accurately mapped to the actual photovoltaic panels, providing a precise spatial reference for subsequent robot cleaning path planning, task allocation, etc. It is the key link between virtual decision-making and physical execution, and ensures the accuracy and effectiveness of multi-robot collaborative cleaning control.

[0089] The identifying of the actual snow area and determining the corresponding snow characteristic parameters by the snow detection model specifically includes:

[0090] Setting a snow pixel value interval, and extracting pixel points in the photovoltaic surface image whose pixel values ​​are within the snow pixel value interval as snow pixel points through the snow detection model;

[0091] Clustering the adjacent snow pixels into a snow pixel set, and determining a snow boundary contour of the actual snow area based on the snow pixel set;

[0092] Calculating the snow accumulation area, snow accumulation area perimeter, and snow distribution range of the actual snow accumulation area based on the snow accumulation boundary contour as basic snow accumulation characteristic parameters;

[0093] Collecting snow distance data of the actual snow area through a lidar sensor, and calculating the average snow thickness and snow thickness distribution gradient in the actual snow area based on the snow distance data as core snow characteristic parameters;

[0094] Summarizing the basic snow characteristic parameters and the core snow characteristic parameters to form the snow characteristic parameters;

[0095] The snow thickness distribution gradient is used to reflect the changing trend of snow thickness in the actual snow area.

[0096] It can be understood that the snow pixel value interval refers to the pixel value range set based on the optical properties of the snow. By setting the snow pixel value interval, snow pixels can be screened out from the photovoltaic surface image. Snow pixels refer to points in the photovoltaic surface image whose pixel values ​​are within the snow pixel value interval, which represent locations where snow may exist. The snow pixel point set refers to a continuous area composed of adjacent snow pixel points, which is used to reflect the distribution pattern of snow in the photovoltaic surface image. The snow boundary contour refers to a closed line formed around the edge of the snow pixel point set, which is used to define the spatial range of the actual snow area.

[0097] Basic snow characteristic parameters describe the geometry of snow areas, including their area, perimeter, and distribution range. The area of ​​the snow area refers to the size of the region enclosed by the snow boundary, reflecting the breadth of snow cover. The perimeter of the snow area refers to the total length of the snow boundary, reflecting the complexity of the snow area. The snow distribution range refers to the location and extent of the snow area on the photovoltaic panel surface.

[0098] A lidar sensor is a device that measures distance by emitting and receiving laser signals and is used to obtain depth information about snow-covered areas. Snow distance data refers to the distance between the snow surface and the sensor, collected by the lidar sensor, and is used to calculate snow thickness. Core snow characteristic parameters describe the physical properties of snow, including average snow thickness and snow thickness distribution gradient. Average snow thickness refers to the average thickness of each point within the snow-covered area and is used to reflect the overall thickness of the snow. The snow thickness distribution gradient refers to a parameter that reflects the rate of change of snow thickness at different locations within the snow-covered area and is used to reflect differences in the spatial distribution of snow thickness.

[0099] First, a pixel value interval that conforms to the optical characteristics of snow can be set. The pixel points in this interval can be extracted from the photovoltaic surface image through the snow detection model and marked as snow pixels, thereby achieving preliminary screening of snow areas.

[0100] Next, the selected snow pixels can be clustered, grouping spatially adjacent pixels into clusters. This process then outlines the actual snow-covered area's boundaries and clarifies the spatial extent of the snow. Based on this outline, the area, perimeter, and distribution of the snow-covered region can be calculated, forming basic snow characteristic parameters that describe the distribution of snow from a geometric perspective. Furthermore, distance data from the actual snow-covered area can be collected using a lidar sensor. This data can then be converted into snow thickness information through data processing. The average snow thickness and snow thickness distribution gradient can then be derived as core snow characteristic parameters to reflect the physical properties of the snow and its thickness trends.

[0101] Finally, the basic characteristic parameters and core characteristic parameters can be summarized and integrated to form complete snow characteristic parameters, which comprehensively cover key information such as the distribution pattern and thickness characteristics of snow, and provide scientific data support for subsequent cleaning mode determination, path planning and parameter adjustment, ensuring the pertinence and effectiveness of the multi-robot collaborative cleaning strategy.

[0102] The determining, based on the snow characteristic parameters and the real-time meteorological data, whether to trigger the robot's extreme snow clearing mode specifically includes:

[0103] Setting the judgment conditions for the extreme snow clearing mode, including a critical value for snow thickness, a critical value for snowfall, a critical value for wind speed, and a critical value for low temperature duration;

[0104] Comparing the average snow thickness in the snow characteristic parameter with the snow thickness critical value, and comparing the hourly snowfall, real-time wind speed, and real-time low temperature duration in the real-time meteorological data with the snowfall critical value, the wind speed critical value, and the low temperature duration critical value, respectively;

[0105] If the average snow thickness exceeds the snow thickness threshold, the hourly snowfall exceeds the snowfall threshold, the real-time wind speed exceeds the preset wind speed threshold, and the real-time low temperature duration exceeds the low temperature duration threshold, then the extreme snow clearing mode of the robot is triggered; otherwise, the normal snow clearing mode of the robot is triggered.

[0106] The snow clearing mode determination result is recorded and associated with the corresponding snow clearing strategy library, and the snow clearing strategy library includes the working parameter configuration and cooperation rules of the robot under the corresponding snow clearing mode.

[0107] The snow thickness threshold is the threshold used to distinguish between normal and extreme snow thicknesses. Exceeding this value indicates that the snow thickness has seriously affected the power generation efficiency of photovoltaic panels. The snowfall amount threshold is the threshold used to measure the intensity of snowfall. Exceeding this value indicates that the snowfall speed may cause rapid snow accumulation. The wind speed threshold is the threshold used to reflect wind strength. Exceeding this value may aggravate snow accumulation or affect the stability of the robot. The low temperature duration threshold is the threshold used to measure the duration of low temperature environment. Exceeding this value may cause snow to freeze, increasing the difficulty of snow removal.

[0108] Average snow thickness refers to the indicator within the snow characteristic parameters that reflects the overall snow thickness. It represents the average thickness of each point in the snow-covered area. Hourly snowfall refers to the accumulated snowfall per unit time and reflects snowfall intensity. Real-time wind speed refers to the current air velocity, which affects snow distribution and robot operation stability. Real-time low temperature duration refers to the duration of the current low temperature environment, which affects the physical state of the snow.

[0109] In addition, the conventional snow clearing mode is suitable for clearing scenes with thin snow depth and stable weather conditions. The snow clearing strategy library is a database used to store operating parameters and collaboration rules for different clearing modes, which can provide standardized operating instructions for robots.

[0110] First, the judgment conditions for extreme snow clearing mode can be preset, covering the critical values ​​of snow thickness, snowfall, wind speed and low temperature duration. These critical values ​​serve as quantitative standards for distinguishing between normal and extreme environments.

[0111] Subsequently, the average snow thickness in the snow characteristic parameters can be compared with the preset snow thickness threshold. At the same time, the hourly snowfall, real-time wind speed, and real-time low temperature duration in the real-time meteorological data can be compared one by one with the corresponding snowfall threshold, wind speed threshold, and low temperature duration threshold. If the results of the above four comparisons all meet the conditions that the actual monitored values ​​exceed the thresholds, that is, the average snow thickness, hourly snowfall, real-time wind speed, and real-time low temperature duration all exceed the corresponding thresholds, then the extreme snow clearing mode is triggered; otherwise, the normal snow clearing mode is triggered.

[0112] Finally, the snow-clearing mode determination results can be recorded and associated with a pre-set snow-clearing strategy library. This strategy library contains the robot's operating parameter configurations and collaboration rules for different modes, providing standardized operational guidance for subsequent robot collaborative operations. This ensures that multiple robots can efficiently complete cleaning tasks according to the adapted strategy in both extreme and normal environments.

[0113] The above process realizes the intelligent switching of snow clearing mode through multi-parameter collaborative judgment and linkage with the strategy library, and improves the robot's adaptability to complex environments.

[0114] The dividing the actual photovoltaic station into a plurality of cleaning sub-areas based on the virtual photovoltaic space and allocating the cleaning sub-areas to the plurality of robots specifically includes:

[0115] The virtual photovoltaic space is marked with arrangement information of an actual photovoltaic panel array;

[0116] Dividing the virtual photovoltaic space into grids according to the arrangement direction of the actual photovoltaic panel array to form initial grid units;

[0117] Marking the initial grid cells covered with snow as cells to be cleaned according to the projection position of the actual snow area in the virtual photovoltaic space;

[0118] According to the preset division direction, that is, the direction from the upper edge to the lower edge of the actual photovoltaic panel, the virtual photovoltaic space is traversed, the total number of pixels in each row and the fraction of pixels in each row belonging to the unit to be cleaned are counted, and the ratio of the fraction to the total number is calculated to obtain the ratio of cleaned pixels;

[0119] When the proportions of the cleaned pixels in a plurality of consecutive rows are all greater than or equal to a preset proportion threshold, a horizontal dividing line is generated, and the generation operation of the horizontal dividing line is repeated until the virtual photovoltaic space is divided into a plurality of longitudinally arranged cleaning sub-areas;

[0120] A safety zone is provided between adjacent cleaning sub-areas, and the width of the safety zone is greater than or equal to the width of the robot body;

[0121] The cleaning sub-areas are merged or split according to the distribution density of the actual snow area until the area difference between any two cleaning sub-areas is within a preset area difference range.

[0122] It should be noted that the actual photovoltaic panel array refers to a collection of multiple actual photovoltaic panels arranged in a preset spacing and arrangement. The arrangement directly affects the planning of the cleaning path. The initial grid cell refers to the basic spatial unit formed after grid division. Each cell corresponds to a specific local area of ​​the actual photovoltaic station and is the basic unit for subsequent marking of the area to be cleaned. The unit to be cleaned refers to the initial grid cell in the virtual photovoltaic space that is marked as having snow and requiring cleaning operations. It is the basis for the subsequent division of cleaning sub-areas.

[0123] The preset division direction refers to the pre-set area division direction, which here is the direction from the upper edge to the lower edge of the actual photovoltaic panel, which is adapted to the arrangement of the photovoltaic panels and the convenience of cleaning operations. The cleaning pixel ratio refers to the ratio of the number of pixels in each row of pixels belonging to the units to be cleaned to the total number of pixels in the row, which is used to reflect the density of snow cover in the row. The preset ratio threshold refers to the critical value used to determine whether to generate a horizontal dividing line. When the cleaning pixel ratio of multiple consecutive rows reaches this value, it indicates that the area is suitable for division into independent cleaning sub-areas. The horizontal dividing line refers to the line generated horizontally in the virtual photovoltaic space, which is used to separate different cleaning sub-areas to ensure the independence of each cleaning sub-area. The cleaning sub-area refers to an independent operating area formed by dividing the horizontal dividing line. Each cleaning sub-area contains a certain number of units to be cleaned.

[0124] A safety zone is a buffer zone between adjacent cleaning sub-areas, used to isolate the robot's operating range in different areas to prevent collisions. The body width refers to the lateral dimension of the robot itself and is an important factor in determining the width of the safety zone, ensuring it is sufficient to accommodate the robot's steering and evasive maneuvers. The preset area difference range is a critical range used to standardize the area consistency of each cleaning sub-area. This ensures a relatively balanced workload across each cleaning sub-area and facilitates the proper distribution of the robot load.

[0125] First, the virtual photovoltaic space can be gridded to form initial grid cells according to the actual arrangement of the photovoltaic panel array, so that the management units of the virtual space are adapted to the distribution characteristics of the actual photovoltaic panels. Then, based on the projection of the actual snow-covered area in the virtual space, the initial grid cells covered with snow can be marked as cells to be cleaned, accurately targeting the areas that need cleaning.

[0126] It should be noted that the division process follows the preset direction, that is, the virtual space is traversed pixel by pixel along the direction from the upper edge to the lower edge of the actual photovoltaic panel, and the pixel ratio of the unit to be cleaned in each row is counted. When the ratio of multiple consecutive rows reaches the preset threshold, a horizontal division line is generated, and the operation is repeated until multiple vertically arranged cleaning sub-areas are formed.

[0127] Furthermore, to ensure the safety of the robots, safety zones are set up between adjacent sub-areas, with a width no less than the width of the robot itself. Finally, the cleaning sub-areas can be merged or split based on the actual snow density, ensuring that the difference in area between any two sub-areas falls within a preset range. This ensures balanced task distribution and lays the foundation for efficient multi-robot collaborative operations.

[0128] Real-time collection of the robot's current status information, including current remaining power, current location coordinates, current workload rate, historical cleaning success rate, and current device health level;

[0129] constructing a snow-clearing capability evaluation score for the robot based on the current state information, wherein the snow-clearing capability evaluation score is used to quantitatively evaluate the current snow-clearing capability of the robot, and the snow-clearing capability evaluation score is obtained by weighting the current battery sufficiency score, the current position adaptability score, the current load bearing capacity score, and the snow-clearing efficiency stability score of the robot;

[0130] Calculating snow clearing priority scores for the cleaning sub-areas, and prioritizing the cleaning sub-areas from largest to smallest according to the snow clearing priority scores, wherein the snow clearing priority scores are weighted by the snow thickness ranking scores, the area ranking scores, the photovoltaic panel power generation capacity scores, and the distance ranking scores between the sub-areas and the robot;

[0131] According to the snow clearing capability evaluation score and the snow clearing priority score, the cleaning sub-area is assigned to the corresponding robot using a dynamic matching rule, and an allocation comparison table of the cleaning sub-areas and the robots is generated and synchronized to the virtual photovoltaic space for visual marking, wherein the marking content includes the current allocation status and the estimated cleaning completion time;

[0132] The snow clearing load rates of the robots are balanced during the allocation process until the difference between the snow clearing load rates of any two robots is within a preset balance difference range.

[0133] It is understood that current status information refers to a set of parameters that reflect the real-time operating status of the robot and is used to determine the robot's operational capabilities and adaptability. The current remaining battery power refers to the remaining energy in the robot's current battery. It is used to determine the length of time the robot can operate sustainably and is a key constraint for task allocation. The current position coordinates refer to the robot's real-time spatial position in the actual photovoltaic field. It is obtained through the positioning system and affects the distance and time it takes for the robot to reach the target area. The current operating load rate refers to the ratio of the number of tasks currently undertaken by the robot to its maximum load capacity, which reflects the saturation level of the robot's current work. The historical cleaning success rate refers to the successful proportion of cleaning tasks completed by the robot in the past, which reflects the reliability and efficiency stability of the robot's operations. The current equipment health level refers to the health level obtained based on the operating status assessment of each component of the robot and is used to reflect the risk of equipment failure.

[0134] Furthermore, the snow clearing capability assessment score refers to a comprehensive indicator used to quantitatively measure the robot's current ability to undertake snow clearing tasks. The higher the score, the stronger the adaptability. The current battery sufficiency score refers to the score calculated based on the current remaining battery power, which is used to reflect the robot's energy guarantee capability for continuous operation. The current position adaptability score refers to the score calculated based on the distance between the current position coordinates and the target sub-area. The closer the distance, the higher the score, which reflects the robot's operational accessibility. The current load bearing capacity score refers to the score calculated based on the current operational load rate. The lower the load rate, the higher the score, which reflects the robot's potential to undertake new tasks. The snow clearing efficiency stability score refers to the score calculated based on the historical clearing success rate, which is used to reflect the stability of the robot's operating efficiency in a snowy environment.

[0135] The snow-clearing priority score is a comprehensive indicator used to quantitatively measure the urgency and importance of clearing a sub-area. A higher score indicates a higher clearing priority. The snow thickness ranking score is a score calculated based on the ranking of the snow thickness of a sub-area among all areas. The thicker the snow thickness, the higher the ranking and the higher the score. The regional area ranking score is a score calculated based on the ranking of the sub-area area among all areas. The larger the area, the higher the ranking and the higher the score. The photovoltaic panel power generation capacity score is a score calculated based on the ranking of the photovoltaic panel power generation capacity of a sub-area among all areas. The stronger the power generation capacity, the higher the score, which reflects the economic value of clearing. The region-to-robot distance ranking score is a score calculated based on the distance ranking between the sub-area and the average position of the robot cluster. The closer the distance, the higher the ranking and the higher the score.

[0136] First, the robot's current status information can be collected in real time, covering key data such as the current remaining power, current position coordinates, current operating load rate, historical cleaning success rate, and current equipment health level. Based on this real-time status information, a snow cleaning capability evaluation score can be constructed to quantitatively evaluate the robot's current snow cleaning capability. This score is calculated by weighting the robot's current power adequacy score, current position adaptability score, current load bearing capacity score, and snow cleaning efficiency stability score. At the same time, the snow cleaning priority score of the cleaning sub-area can be calculated. This score is obtained by weighting the snow thickness ranking score, area ranking score, photovoltaic panel power generation capacity score, and distance ranking score of the cleaning sub-area. The cleaning sub-areas are then prioritized in descending order according to the priority score.

[0137] It should be noted that dynamic matching rules refer to rules that adjust allocation relationships in real time based on robot capabilities and the priorities of cleaning sub-areas. Furthermore, dynamic matching rules can be used to assign cleaning sub-areas to corresponding robots based on the snow-clearing capability assessment score and the snow-clearing priority score. This generates a comparison table of cleaning sub-areas and robot allocations and synchronizes them to the virtual photovoltaic space for visual marking. The markings include the current allocation status and the estimated cleaning completion time. During the allocation process, attention is paid to balancing the robots' snow-clearing load rates until the difference in snow-clearing load rates between any two robots falls within a preset equilibrium difference range, thereby achieving efficient and balanced multi-robot collaborative operations.

[0138] The process of determining the snow clearing path of the robot is as follows:

[0139] generating an initial straight path for the robot based on a virtual starting position of the robot in the virtual photovoltaic space and a virtual arrival position corresponding to the cleaning sub-area;

[0140] calling an obstacle database pre-stored in the virtual photovoltaic space to perform obstacle detection on the initial straight path, wherein the obstacle database includes three-dimensional coordinate information of static obstacles;

[0141] If there is an obstacle in the initial straight path, a multi-segment broken line obstacle avoidance path is generated using the obstacle avoidance rule and the initial straight path is adjusted to generate the snow clearing path until the distance between the snow clearing path and the obstacle is greater than or equal to the preset safety distance.

[0142] The virtual starting position refers to the robot's initial position mapping in the virtual photovoltaic space. This coordinate system aligns with the actual starting position and serves as the starting point for path planning. The virtual arrival position refers to the target location of the cleaning subarea in the virtual photovoltaic space, typically the geometric center or entrance of the cleaning subarea, serving as the endpoint for path planning. The initial straight path, a straight line connecting the virtual starting position and the virtual arrival position, forms the basis for path planning.

[0143] The obstacle database is a collection of obstacle information stored in the virtual PV space. It contains the spatial location data of various static obstacles, providing a basis for obstacle avoidance in path planning. Static obstacles are fixed obstacles in the actual PV site, such as PV racks, cables, substations, and columns, whose positions do not change over time. Three-dimensional coordinate information describes the spatial location of obstacles, including horizontal, vertical, and height information. This information allows precise positioning of the obstacle's spatial extent.

[0144] Obstacle avoidance rules refer to a set of strategies used to plan obstacle avoidance paths. These include principles such as circumventing obstacle edges and calculating the shortest detour distance, ensuring that the generated obstacle avoidance paths are feasible and efficient. A multi-segment broken-line obstacle avoidance path is a detour formed by connecting multiple straight lines. It avoids obstacles by changing direction and is easier for the robot to control than a curved path. The preset safety distance refers to the minimum spacing standard set to prevent collisions between the robot and obstacles. It needs to be determined based on the robot's size, motion accuracy, and obstacle type to ensure sufficient safety buffer space.

[0145] According to the snow thickness distribution gradient in the snow characteristic parameter, a path segment in the snow clearing path where the snow thickness distribution gradient exceeds a preset gradient threshold is marked as a key snow clearing path segment, and the key snow clearing path segment accounts for greater than or equal to 60% of the path length of the snow clearing path;

[0146] The robots share the snow-clearing paths through wireless communication. If it is detected that the snow-clearing paths of different robots overlap in the same time period, the overlapping paths of the different snow-clearing paths need to be adjusted until there is no overlapping path in the same time period.

[0147] The preset gradient threshold is the critical value used to define a critical path segment. Path segments exceeding this threshold require focused cleaning due to drastic changes in snow thickness. A critical snow-clearing path segment is a portion of a snow-clearing path where the snow thickness distribution gradient exceeds the preset threshold. These areas are key cleaning areas and require enhanced cleaning strategies. The path length ratio refers to the ratio of the total length of the critical snow-clearing path segment to the total length of the entire snow-clearing path. Setting a minimum ratio ensures that key areas are adequately cleared.

[0148] It should be noted that, based on the snow thickness distribution gradient reflecting the snow thickness change trend in the snow characteristic parameters, the path segments in the snow clearing path where the gradient exceeds the preset gradient threshold can be marked as critical snow clearing path segments, and the length of such critical path segments in the entire snow clearing path shall account for no less than 60%.

[0149] The robots share their respective snow-clearing paths through wireless communication. If it is detected that the paths of different robots overlap in the same time period, these overlapping paths need to be adjusted until the paths of each robot no longer overlap in the same time period, so as to ensure the orderliness and safety of multi-robot collaborative operations.

[0150] The updating process of the virtual photovoltaic space is as follows:

[0151] After the robot completes the snow clearing task in the corresponding cleaning sub-area, it uploads a cleaning completion signal to the control center, and simultaneously monitors the surface cleanliness of the actual photovoltaic panel and transmits the signal to the control center;

[0152] The control center marks the cleaned sub-area in the virtual photovoltaic space as cleaned according to the cleaning completion signal, and the remaining cleaning sub-areas remain in a to-be-cleaned state;

[0153] The control center associates the surface cleanliness with a virtual photovoltaic panel in the virtual photovoltaic space, and visually marks the virtual photovoltaic panel using a color gradient;

[0154] The control center synchronously updates the real-time virtual position of the robot in the virtual photovoltaic space;

[0155] Associating the start time and end time of the robot performing the snow clearing task with the time axis of the virtual photovoltaic space to form a dynamic snow clearing process record;

[0156] The virtual photovoltaic space supports remote access operations of the monitoring terminal.

[0157] The "cleaning completion signal" is a status signal generated by a robot after completing the snow-clearing task in a cleaning sub-area. It notifies the control center that the cleaning operation in that sub-area has concluded. The control center is a core control unit that receives, processes, stores, and issues commands. It is responsible for coordinating the multi-robot operation and virtual space updates.

[0158] The "Cleaned" status is a status tag in the virtual PV space that indicates that a cleaning sub-area has been completed. This status tag typically corresponds precisely to the area that has actually been cleaned. The "To Be Cleaned" status is a status tag in the virtual PV space that indicates that a cleaning sub-area has not yet started or completed, indicating that the area still requires processing.

[0159] Color gradient refers to a visualization method that uses different shades or tones of color to represent differences in data magnitude. For example, dark colors represent low cleanliness and light colors represent high cleanliness, making it easy to quickly identify the cleanliness distribution.

[0160] Real-time virtual position refers to the dynamic mapping position of the robot in the virtual photovoltaic space. It achieves real-time synchronization with the actual position through positioning technology and coordinate transformation, reflecting the instantaneous state of the robot.

[0161] The timeline is a linear marker used to record the chronological order of events in the virtual photovoltaic space. Furthermore, the dynamic snow clearing process record can fully present the execution process and time nodes of the snow clearing task.

[0162] In actual applications, when the robot completes the snow-clearing task in the corresponding cleaning sub-area, it will upload a cleaning completion signal to the control center, and at the same time transmit the monitored surface cleanliness data of the actual photovoltaic panels to the control center, providing a basic basis for updating the virtual space.

[0163] Upon receiving the cleaning completion signal, the control center marks the cleaned sub-areas in the virtual photovoltaic space as cleaned, while unfinished areas remain in the pending cleaning state, clearly displaying the overall cleaning progress. The control center also associates surface cleanliness with virtual photovoltaic panels in the virtual photovoltaic space and visually marks them using a color gradient. This color difference intuitively reflects the cleanliness level of different areas, allowing monitoring terminals to quickly assess cleaning quality.

[0164] The control center also synchronizes the robot's real-time virtual position within the virtual photovoltaic space, ensuring consistency between the virtual location and the robot's actual position. The control center also links the robot's mission start and end times to the virtual space's timeline, creating a dynamic record of the snow-clearing process and enabling a complete traceability of the clearing operation. Furthermore, the virtual photovoltaic space supports remote access from the monitoring end, facilitating remote control and decision-making, enabling comprehensive control of the clearing process.

[0165] Example 2

[0166] like Figure 2 As shown, a multi-robot collaborative photovoltaic panel cleaning control system includes:

[0167] The photovoltaic space retrieval module is used to receive the cleaning task request from the monitoring end and retrieve the virtual photovoltaic space corresponding to the actual photovoltaic station;

[0168] A snow area recognition module is used to collect photovoltaic surface images of actual photovoltaic panels in the actual photovoltaic station and input them into a snow detection model. The snow detection model is used to identify the actual snow area and determine the corresponding snow characteristic parameters. At the same time, real-time meteorological data is obtained through meteorological monitoring equipment.

[0169] a cleaning mode determination module, configured to determine whether to trigger the robot's extreme snow cleaning mode based on the snow characteristic parameters and the real-time meteorological data;

[0170] a cleaning area allocation module, configured to, if the extreme snow cleaning mode of the robot is triggered, divide the actual photovoltaic station into a plurality of cleaning sub-areas based on the virtual photovoltaic space and allocate the sub-areas to the plurality of robots, and simultaneously determine the snow cleaning paths and snow cleaning parameters of the robots;

[0171] a cleaning task execution module, configured for the robot to perform the snow cleaning task according to the snow cleaning path and the snow cleaning parameters, while monitoring the actual surface cleanliness of the photovoltaic panel;

[0172] A photovoltaic space update module is used to adjust the snow removal parameters and update the virtual photovoltaic space based on the surface cleanliness.

[0173] Through the introduction of the above embodiments, the present invention uses a multi-robot collaborative photovoltaic panel cleaning control method and system, which receives a cleaning task request from a monitoring terminal and retrieves a virtual photovoltaic space corresponding to an actual photovoltaic station; collects photovoltaic surface images of actual photovoltaic panels in the actual photovoltaic station and inputs them into a snow detection model, identifies the actual snow area through the snow detection model and determines the corresponding snow characteristic parameters, and simultaneously obtains real-time meteorological data through meteorological monitoring equipment; determines whether to trigger the robot's extreme snow cleaning mode based on the snow characteristic parameters and real-time meteorological data; if the robot's extreme snow cleaning mode is triggered, the actual photovoltaic station is divided into multiple cleaning sub-areas based on the virtual photovoltaic space and assigned to multiple robots, and the robots' snow cleaning paths and snow cleaning parameters are determined; the robots perform snow cleaning tasks according to the snow cleaning paths and snow cleaning parameters, while monitoring the surface cleanliness of the actual photovoltaic panels; based on the surface cleanliness, the snow cleaning parameters are adjusted and the virtual photovoltaic space is updated, thereby improving the efficiency of multi-robot collaborative cleaning of photovoltaic panels in extreme weather and achieving precise adjustment and optimization of the cleaning mode.

[0174] The present invention achieves intelligent triggering of extreme snow-clearing mode by constructing a precise mapping between virtual photovoltaic space and actual photovoltaic stations, combining snow detection models with real-time meteorological data, solving the problem of low efficiency of multi-robot collaboration in extreme weather conditions and significantly improving the response speed of photovoltaic panel cleaning. The method of the present invention is based on a task allocation mechanism based on dynamic division of sub-areas in virtual space and capability matching, which balances the robot load. Combined with the obstacle avoidance and anti-overlap design in path planning, it reduces ineffective robot operations and improves the overall snow-clearing efficiency. The present invention dynamically adjusts the robot's snow-clearing parameters based on snow characteristic parameters, and relies on virtual space to update the snow-clearing status in real time, achieving refined control of the snow-clearing process, significantly improving the compliance rate of photovoltaic panel cleanliness, and ensuring normal power generation of photovoltaic panels. By integrating dynamic adjustment and collaborative mechanisms, the present invention enhances the adaptability of multiple robots to complex snow environments, reduces equipment losses, and reduces the cost of manual intervention.

[0175] 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 box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes 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 produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the steps in the process. 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.

[0176] Those skilled in the art will appreciate that all or part of the steps in the various methods of the above embodiments can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium, including a read-only memory (ROM), a random access memory (RAM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), a one-time programmable read-only memory (OTPROM), an electronically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disc storage, magnetic disk storage, magnetic tape storage, or any other computer-readable medium capable of carrying or storing data.

[0177] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a series of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, commodity, or apparatus that includes the element.

Claims

1. A multi-robot collaborative photovoltaic panel cleaning control method, characterized in that: The method comprises: Receive cleaning task requests from the monitoring end and retrieve the virtual photovoltaic space corresponding to the actual photovoltaic site; Collecting photovoltaic surface images of actual photovoltaic panels in the actual photovoltaic station and inputting them into a snow detection model, identifying actual snow areas and determining corresponding snow characteristic parameters through the snow detection model, and simultaneously obtaining real-time meteorological data through meteorological monitoring equipment; determining whether to trigger an extreme snow clearing mode of the robot based on the snow characteristic parameters and the real-time meteorological data; If the extreme snow clearing mode of the robot is triggered, the actual photovoltaic station is divided into a plurality of clearing sub-areas based on the virtual photovoltaic space and the sub-areas are assigned to a plurality of robots, and the snow clearing paths and snow clearing parameters of the robots are determined at the same time; The robot performs the snow clearing task according to the snow clearing path and the snow clearing parameters, while monitoring the surface cleanliness of the actual photovoltaic panel; Based on the surface cleanliness, the snow removal parameters are adjusted and the virtual photovoltaic space is updated.

2. A multi-robot collaborative photovoltaic panel cleaning control method according to claim 1, characterized in that: The virtual photovoltaic space includes a virtual photovoltaic panel corresponding to the actual photovoltaic panel, and a coordinate system of the virtual photovoltaic panel is the same as that of the actual photovoltaic panel.

3. A multi-robot collaborative photovoltaic panel cleaning control method according to claim 1, characterized in that: The identifying of the actual snow area and determining the corresponding snow characteristic parameters by the snow detection model specifically includes: Setting a snow pixel value interval, and extracting pixel points in the photovoltaic surface image whose pixel values ​​are within the snow pixel value interval as snow pixel points through the snow detection model; Clustering the adjacent snow pixels into a snow pixel set, and determining a snow boundary contour of the actual snow area based on the snow pixel set; Calculating the snow accumulation area, snow accumulation area perimeter, and snow distribution range of the actual snow accumulation area based on the snow accumulation boundary contour as basic snow accumulation characteristic parameters; Collecting snow distance data of the actual snow area through a lidar sensor, and calculating the average snow thickness and snow thickness distribution gradient in the actual snow area based on the snow distance data as core snow characteristic parameters; Summarizing the basic snow characteristic parameters and the core snow characteristic parameters to form the snow characteristic parameters; The snow thickness distribution gradient is used to reflect the changing trend of snow thickness in the actual snow area.

4. A multi-robot collaborative photovoltaic panel cleaning control method according to claim 1, characterized in that: The determining, based on the snow characteristic parameters and the real-time meteorological data, whether to trigger the robot's extreme snow clearing mode specifically includes: Setting the judgment conditions for the extreme snow clearing mode, including a critical value for snow thickness, a critical value for snowfall, a critical value for wind speed, and a critical value for low temperature duration; Comparing the average snow thickness in the snow characteristic parameter with the snow thickness critical value, and comparing the hourly snowfall, real-time wind speed, and real-time low temperature duration in the real-time meteorological data with the snowfall critical value, the wind speed critical value, and the low temperature duration critical value, respectively; If the average snow thickness exceeds the snow thickness threshold, the hourly snowfall exceeds the snowfall threshold, the real-time wind speed exceeds the wind speed threshold, and the real-time low temperature duration exceeds the low temperature duration threshold, then the extreme snow clearing mode of the robot is triggered; otherwise, the normal snow clearing mode of the robot is triggered. The snow clearing mode determination result is recorded and associated with the corresponding snow clearing strategy library, and the snow clearing strategy library includes the working parameter configuration and cooperation rules of the robot under the corresponding snow clearing mode.

5. The multi-robot collaborative photovoltaic panel cleaning control method according to claim 1, characterized in that: The dividing the actual photovoltaic station into a plurality of cleaning sub-areas based on the virtual photovoltaic space and allocating the cleaning sub-areas to the plurality of robots specifically includes: The virtual photovoltaic space is marked with arrangement information of an actual photovoltaic panel array; Dividing the virtual photovoltaic space into grids according to the arrangement direction of the actual photovoltaic panel array to form initial grid units; Marking the initial grid cells covered with snow as cells to be cleaned according to the projection position of the actual snow area in the virtual photovoltaic space; According to the preset division direction, that is, the direction from the upper edge to the lower edge of the actual photovoltaic panel, the virtual photovoltaic space is traversed, the total number of pixels in each row and the fraction of pixels in each row belonging to the unit to be cleaned are counted, and the ratio of the fraction to the total number is calculated to obtain the ratio of cleaned pixels; When the proportions of the cleaned pixels in a plurality of consecutive rows are all greater than or equal to a preset proportion threshold, a horizontal dividing line is generated, and the generation operation of the horizontal dividing line is repeated until the virtual photovoltaic space is divided into a plurality of longitudinally arranged cleaning sub-areas; A safety zone is provided between adjacent cleaning sub-areas, and the width of the safety zone is greater than or equal to the width of the robot body; The cleaning sub-areas are merged or split according to the distribution density of the actual snow area until the area difference between any two cleaning sub-areas is within a preset area difference range.

6. A multi-robot collaborative photovoltaic panel cleaning control method according to claim 5, characterized in that: Real-time collection of the robot's current status information, including current remaining power, current location coordinates, current workload rate, historical cleaning success rate, and current device health level; constructing a snow-clearing capability evaluation score for the robot based on the current state information, wherein the snow-clearing capability evaluation score is used to quantitatively evaluate the current snow-clearing capability of the robot, and the snow-clearing capability evaluation score is obtained by weighting the current battery sufficiency score, the current position adaptability score, the current load bearing capacity score, and the snow-clearing efficiency stability score of the robot; Calculating snow clearing priority scores for the cleaning sub-areas, and prioritizing the cleaning sub-areas from largest to smallest according to the snow clearing priority scores, wherein the snow clearing priority scores are weighted by the snow thickness ranking scores, the area ranking scores, the photovoltaic panel power generation capacity scores, and the distance ranking scores between the sub-areas and the robot; According to the snow clearing capability evaluation score and the snow clearing priority score, the cleaning sub-area is assigned to the corresponding robot using a dynamic matching rule, and an allocation comparison table of the cleaning sub-areas and the robots is generated and synchronized to the virtual photovoltaic space for visual marking, wherein the marking content includes the current allocation status and the estimated cleaning completion time; The snow clearing load rates of the robots are balanced during the allocation process until the difference between the snow clearing load rates of any two robots is within a preset balance difference range.

7. The multi-robot collaborative photovoltaic panel cleaning control method according to claim 1, characterized in that: The process of determining the snow clearing path of the robot is as follows: generating an initial straight path for the robot based on a virtual starting position of the robot in the virtual photovoltaic space and a virtual arrival position corresponding to the cleaning sub-area; calling an obstacle database pre-stored in the virtual photovoltaic space to perform obstacle detection on the initial straight path, wherein the obstacle database includes three-dimensional coordinate information of static obstacles; If there is an obstacle in the initial straight path, a multi-segment broken line obstacle avoidance path is generated using the obstacle avoidance rule and the initial straight path is adjusted to generate the snow clearing path until the distance between the snow clearing path and the obstacle is greater than or equal to the preset safety distance.

8. A multi-robot collaborative photovoltaic panel cleaning control method according to claim 7, characterized in that: According to the snow thickness distribution gradient in the snow characteristic parameter, a path segment in the snow clearing path where the snow thickness distribution gradient exceeds a preset gradient threshold is marked as a key snow clearing path segment, and the key snow clearing path segment accounts for greater than or equal to 60% of the path length of the snow clearing path; The robots share the snow-clearing paths through wireless communication. If it is detected that the snow-clearing paths of different robots overlap in the same time period, the overlapping paths of the different snow-clearing paths need to be adjusted until there is no overlapping path in the same time period.

9. The multi-robot collaborative photovoltaic panel cleaning control method according to claim 1, characterized in that: The updating process of the virtual photovoltaic space is as follows: After the robot completes the snow clearing task in the corresponding cleaning sub-area, it uploads a cleaning completion signal to the control center, and simultaneously monitors the surface cleanliness of the actual photovoltaic panel and transmits the signal to the control center; The control center marks the cleaned sub-area in the virtual photovoltaic space as cleaned according to the cleaning completion signal, and the remaining cleaning sub-areas remain in a to-be-cleaned state; The control center associates the surface cleanliness with a virtual photovoltaic panel in the virtual photovoltaic space, and visually marks the virtual photovoltaic panel using a color gradient; The control center synchronously updates the real-time virtual position of the robot in the virtual photovoltaic space; Associating the start time and end time of the robot performing the snow clearing task with the time axis of the virtual photovoltaic space to form a dynamic snow clearing process record; The virtual photovoltaic space supports remote access operations of the monitoring terminal.

10. A multi-robot collaborative photovoltaic panel cleaning control system, applied to a multi-robot collaborative photovoltaic panel cleaning control method according to any one of claims 1 to 9, characterized in that: The system comprises: The photovoltaic space retrieval module is used to receive the cleaning task request from the monitoring end and retrieve the virtual photovoltaic space corresponding to the actual photovoltaic station; A snow area recognition module is used to collect photovoltaic surface images of actual photovoltaic panels in the actual photovoltaic station and input them into a snow detection model. The snow detection model is used to identify the actual snow area and determine the corresponding snow characteristic parameters. At the same time, real-time meteorological data is obtained through meteorological monitoring equipment. a cleaning mode determination module, configured to determine whether to trigger the robot's extreme snow cleaning mode based on the snow characteristic parameters and the real-time meteorological data; a cleaning area allocation module, configured to, if the extreme snow cleaning mode of the robot is triggered, divide the actual photovoltaic station into a plurality of cleaning sub-areas based on the virtual photovoltaic space and allocate the sub-areas to the plurality of robots, and simultaneously determine the snow cleaning paths and snow cleaning parameters of the robots; a cleaning task execution module, configured for the robot to perform the snow cleaning task according to the snow cleaning path and the snow cleaning parameters, while monitoring the actual surface cleanliness of the photovoltaic panel; A photovoltaic space update module is used to adjust the snow removal parameters and update the virtual photovoltaic space based on the surface cleanliness.

Images