An intelligent system for photovoltaic cleaning robots

By integrating intelligent control center and machine learning technology in photovoltaic cleaning robots, automatic inspection and cleaning are achieved, and the problems of inconvenient operation, low efficiency and high cost of existing photovoltaic cleaning robots are solved, and unmanned intelligent operation and maintenance are achieved, improving Cleaning efficiency and convenience.

CN117593675BActive Publication Date: 2024-11-12INNER MONGOLIA UNIV OF TECH
View PDF 0 Cites 1 Cited by

Patent Information

Application Number
CN202410073949.8
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-01-18
Publication Date
2024-11-12
Estimated Expiration
2044-01-18

AI Technical Summary

Technical Problem

Existing photovoltaic cleaning robots are inconvenient to operate, have low cleaning efficiency, high cost, and are difficult to achieve timely cleaning and maintenance in outdoor environments.

Method used

Design a photovoltaic cleaning robot system including control terminals, intelligent control centers, stain detection databases and intelligent devices, using camera shooting modules, motion flight modules and cleaning modules, combining machine learning technology and wireless communications to realize automatic patrol inspections and cleaning, the body comes with sensor to monitor the status, optimize the cleaning route and execute instructions.

Benefits of technology

It realizes unmanned intelligent operation and maintenance, reduces operation and maintenance costs, improves cleaning efficiency and effect, ensures timely cleaning of photovoltaic panels, simplifies the operation process, and improves the convenience and safety of the system.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN117593675B_ABST
    Figure CN117593675B_ABST
Patent Text Reader

Abstract

The present invention discloses an intelligent system of a photovoltaic cleaning robot, which belongs to the technical field of photovoltaic cleaning robots, and includes a control terminal, an intelligent control center, a stain detection database and an intelligent device; the intelligent device includes a camera shooting module, a motion flight module and a cleaning module; the intelligent device and the intelligent control center are integrated on the photovoltaic cleaning robot; the intelligent control center is connected to the control terminal through wireless communication; the intelligent control center is connected to the stain detection database through wireless communication. The present invention provides an intelligent system of a photovoltaic cleaning robot, which can achieve unmanned intelligent cleaning and operation and maintenance of photovoltaic panels even in distributed photovoltaic power stations in harsh environments and terrains. The operator only needs to operate the control terminal to issue simple instructions according to the attenuation of the power generation of the power station, and the photovoltaic cleaning robot can perform full-process intelligent operation and maintenance, which greatly improves the convenience.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the technical field of photovoltaic cleaning robots, and in particular relates to an intelligent system of a photovoltaic cleaning robot. Background Art

[0002] As a green and sustainable energy, solar energy plays an important role in long-term energy strategies. In photovoltaic power stations built in the wild with complex terrain, photovoltaic modules are often arranged in a distributed manner. Due to various factors such as dry climate, strong wind and sand, complex terrain, and difficult maintenance, a large amount of dust and other covering materials are often deposited on the surface of photovoltaic modules, causing a serious decrease in power generation efficiency, and even causing photovoltaic panels to produce hot spot effects, directly damaging the life of photovoltaic modules. Photovoltaic cleaning robots that can achieve high-pressure flushing by adding water tanks and high-pressure water pumps to drone platforms can overcome complex terrain to clean photovoltaic solar panels. However, such photovoltaic cleaning robots on the market now still require professional operators to control the equipment through terminals, which is very inconvenient and has low cleaning efficiency. At the same time, the outdoor environment is harsh and transportation is inconvenient. Each cleaning involves handling equipment and hiring professional operators, resulting in low timeliness in cleaning photovoltaic panels and increasing the cost of cleaning operations. Summary of the invention

[0003] The purpose of the present invention is to provide an intelligent system of a photovoltaic cleaning robot to solve the problems of inconvenient operation, low cleaning efficiency and high cleaning cost when cleaning photovoltaic panels in the prior art.

[0004] To achieve the above-mentioned purpose, the present invention provides an intelligent system of a photovoltaic cleaning robot, comprising a control terminal, an intelligent control center, a stain detection database and an intelligent device; the intelligent device comprises a camera shooting module, a motion flight module and a cleaning module; the intelligent device and the intelligent control center are both integrated on the photovoltaic cleaning robot; the intelligent control center is connected to the control terminal via wireless communication; the intelligent control center is connected to the stain detection database via wireless communication;

[0005] The intelligent control center receives a start command from the control terminal: when the position of the fuselage is detected to be beyond the threshold of the last sleep position, it returns to the control terminal to determine the inspection range; if there is no latitude change, it returns to the control terminal whether to inspect, and if it is not, the automatic cleaning mode is turned on;

[0006] The intelligent control center is connected to the stain detection database in real time, so as to facilitate the continuous updating and learning of the machine learning algorithm and improve the accuracy of detection and cleaning;

[0007] The intelligent control center is connected to the control terminal in real time to transmit data and pictures of the cleaning operation or receive instructions from the control terminal;

[0008] The intelligent control center sends an execution instruction to the photovoltaic cleaning robot to control the photovoltaic cleaning robot to enter an intelligent cleaning mode.

[0009] Preferably, the smart device further comprises:

[0010] The image acquisition module collects the image information of the photovoltaic panel through the onboard camera shooting module. The image information includes the surface stains, damage or abnormality of the photovoltaic panel;

[0011] The image information processing module uses machine learning technology to analyze and identify the collected image information, thereby obtaining the required operating instructions, identifying different types of stains, estimating the severity of stains, and identifying physical damage to photovoltaic panels;

[0012] The body monitoring module provides obstacle avoidance for the photovoltaic cleaning robot through the body's built-in sensors combined with the image information processing module, monitors the body's power and the remaining amount of cleaning agent, and obtains conditional instructions II;

[0013] The instruction judgment module is used to judge the priority of each conditional instruction and give the execution instruction through the intelligent system, including the condition comparison submodule, which is used to judge the priority of the conditional instruction; the instruction output submodule, which intelligently outputs the execution instruction according to the priority of the conditional instruction;

[0014] The control processing module is used to control the photovoltaic cleaning robot to execute instructions.

[0015] Preferably, the specific execution process of the intelligent cleaning mode is as follows:

[0016] S100, the control terminal issues a device start command, and the photovoltaic cleaning robot enters the intelligent cleaning mode;

[0017] S200, establishing a recognition model, automatically inspecting based on the established recognition model, and collecting photovoltaic panel location information;

[0018] S300, automatically planning a cleaning route and verifying learning based on the collected photovoltaic panel location information;

[0019] S400, storing the cleaning route and controlling the photovoltaic cleaning robot to perform cleaning operations;

[0020] S500, according to different working stages, monitor the fuselage status and working status in real time during the operation process to determine whether to return to the supply station.

[0021] Preferably, the specific process of establishing the recognition model in S200 is as follows:

[0022] S21, based on deep learning, uses lightweight GhostNet convolution calculation. The specific calculation formula is as follows:

[0023] ;

[0024] ;

[0025] in, express indivual aisle The convolution kernel size is Indicates A linear calculation is used to generate the Ghost feature map, represents the input feature map, represents the output feature map processed by the convolution layer, express Middle The channel feature map The linear operation results are: Indicates the number of channels, Represents an internal parameter, express Middle Channel feature maps;

[0026] S22. Use the CloU loss function to speed up and complete the model building. The specific expression is as follows:

[0027] ;

[0028] ;

[0029] ;

[0030] ;

[0031] ;

[0032] In the formula, Indicates that the center point of the predicted box and the real box is The distance of the direction, Indicates that the center point of the predicted box and the real box is The distance of the direction, represents the difference in bounding box lengths, represents the original loss function, represents the new improved loss function, and are the height and width of the feature map, respectively. is the regularization coefficient, is the standard deviation of the width and length of the prediction box.

[0033] Preferably, in S100, the control terminal issues a device start instruction, and the specific process of the photovoltaic cleaning robot entering the intelligent cleaning mode is as follows:

[0034] S110, the control terminal issues a start command;

[0035] S120, determining the position information of the fuselage and returning an instruction to the terminal whether to inspect;

[0036] S130: If the intelligent system determines the inspection range, the body status and power level are checked and the water tank is drained.

[0037] Preferably, the instructions for determining the fuselage position information and returning to the terminal for inspection in S120 specifically include:

[0038] S121, when the control terminal chooses not to inspect, the stored cleaning route is retrieved according to the location information;

[0039] S122, when the control terminal selects inspection, the inspection range is manually set or the system defaults to the inspection range;

[0040] S123: When it is detected that the position change of the fuselage exceeds the preset range, the inspection range is manually set or the system defaults to the inspection range.

[0041] Preferably, the specific process of automatically inspecting and collecting the location information of the photovoltaic panels in S200 based on the established recognition model is as follows:

[0042] S210, intelligent control center automatically plans inspection routes according to the inspection range;

[0043] S220, flying according to the inspection route, identifies the photovoltaic panels and stores the location information through machine learning technology, and maps it to the regional map.

[0044] Preferably, the specific process of automatically planning the cleaning route and verifying the learning according to the collected photovoltaic panel position information in S300 is as follows:

[0045] S310, automatically planning a cleaning route according to the photovoltaic panel location information and performing an empty-load test flight;

[0046] S320, avoids obstacles and collects obstacle information through body sensors and machine learning technology;

[0047] S330. Replan the cleaning route according to the obstacle information and return the result to the control terminal.

[0048] Preferably, the specific process of storing the cleaning route in S400 to control the photovoltaic cleaning robot to perform the cleaning operation is as follows:

[0049] S410, start intelligent cleaning, and the robot returns to the supply station to charge and add cleaning agent;

[0050] S420, based on the location information of the cleaning route, flying to and identifying the first photovoltaic panel of the route through the intelligent control center and machine learning technology;

[0051] S430, automatically identifying and judging whether the photovoltaic panel needs to be cleaned according to machine learning technology;

[0052] S431, automatically flying to the next photovoltaic panel on the cleaning route when cleaning is not needed, and recording the progress of the cleaning route;

[0053] S432, when cleaning is required, spray cleaning agent, judge the cleaning status of the photovoltaic panel, and after reaching the set cleaning standard, fly to the next photovoltaic panel on the route to continue the cleaning operation;

[0054] S440. When cleaning is completed, return to the supply station and record the progress of the cleaning route.

[0055] Preferably, the specific process of determining whether to return to the supply station in S500 is as follows:

[0056] S510, in the automatic inspection stage, the fuselage status and power are detected. When the fuselage status is abnormal or the power is insufficient, the aircraft returns to the supply station for replenishment. After replenishment, the aircraft automatically plans the shortest distance according to the progress of the operation route, and in combination with the obstacle avoidance system, flies directly from the supply station to the next operation block on the operation route;

[0057] S520: During the intelligent cleaning stage, the fuselage status, power level and remaining cleaning agent are detected. When an abnormal fuselage status or insufficient power is detected, the robot returns to the supply station for replenishment. After replenishment, the robot automatically plans the shortest distance based on the progress of the operation route, and in combination with the obstacle avoidance system, flies directly from the supply station to the next section to be operated on the operation route.

[0058] Therefore, the present invention adopts the above-mentioned intelligent system of a photovoltaic cleaning robot, which has the following beneficial effects:

[0059] 1. In this application, the operator only needs to issue simple instructions on the terminal in combination with the server information of the power station to realize the unmanned photovoltaic panel cleaning and maintenance work throughout the process, which is convenient and fast, and ensures the timeliness of photovoltaic panel cleaning.

[0060] 2. In this application, when the photovoltaic cleaning robot is executing instructions, the self-monitoring module of the fuselage is fully turned on to monitor the working status of the photovoltaic cleaning robot in real time to ensure the smooth and safe execution of operation and maintenance work.

[0061] 3. In this application, the photovoltaic cleaning robot is equipped with a positioning system and an intelligent control system, and uses machine learning technology to connect to the cloud-based photovoltaic panel stain detection database, thereby ensuring the intelligence of the photovoltaic cleaning robot. Even in distributed and complex terrain photovoltaic power stations, unmanned intelligent operation and maintenance can be achieved, thereby reducing operation and maintenance costs and improving the operation and maintenance efficiency of photovoltaic power plants.

[0062] 4. In this application, the photovoltaic cleaning robot uses machine learning technology to identify external scenes, photovoltaic panels and stains, and uses an integrated intelligent control center to plan steps and issue instructions to control the robot's operations. This model eliminates cumbersome operations, saves manpower, and greatly improves the efficiency and effectiveness of photovoltaic operation and maintenance.

[0063] 5. In this application, the photovoltaic cleaning robot will perform positioning detection before operation and maintenance to identify whether re-inspection is required. If re-inspection is not required, the cleaning and maintenance work will be started directly according to the route and coordinates stored in the system, ensuring the smooth progress of photovoltaic panel cleaning and maintenance work, and will not affect the safety of the fuselage due to errors in the new site.

[0064] 6. In this application, the control terminal can issue instructions to the intelligent photovoltaic cleaning robot during operation or in a shutdown state. The intelligent system of the photovoltaic cleaning robot will prioritize the execution of the instructions issued by the terminal based on environmental factors and the state of the body. If the instructions cannot be executed, the terminal will be returned with the reason for non-execution, thereby improving convenience and safety.

[0065] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 It is an overall block diagram of an intelligent system of a photovoltaic cleaning robot of the present invention;

[0067] Figure 2 is a structural block diagram of the intelligent device of the present invention;

[0068] Figure 3 It is a flowchart of the intelligent cleaning mode of the present invention;

[0069] Figure 4 It is a detailed flow chart of the intelligent cleaning mode of the present invention. DETAILED DESCRIPTION

[0070] The following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. 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.

[0071] See also Figure 1-4 , an intelligent system of a photovoltaic cleaning robot, comprising a control terminal, an intelligent control center, a stain detection database and an intelligent device; the intelligent device comprises a camera shooting module, a motion flight module and a cleaning module; the intelligent device and the intelligent control center are both integrated on the photovoltaic cleaning robot; the intelligent control center is connected to the control terminal through wireless communication; the intelligent control center is connected to the stain detection database through wireless communication; the intelligent control center receives a start-up command issued by the control terminal: when it is detected that the position of the fuselage exceeds the threshold of the last sleep position range, it returns to the control terminal to determine the inspection range; if there is no latitude change, it returns to the control terminal whether to inspect, and if it is selected not, the automatic cleaning mode is turned on; the intelligent control center is connected to the stain detection database in real time, so as to facilitate the continuous updating and learning of the machine learning algorithm and improve the accuracy of detection and cleaning; the intelligent control center is connected to the control terminal in real time, and transmits back the data and pictures of the cleaning operation, or receives the command of the control terminal; the intelligent control center sends an execution command to the photovoltaic cleaning robot to control the photovoltaic cleaning robot to enter the intelligent cleaning mode.

[0072] The intelligent device also includes: an image acquisition module, which collects image information of the photovoltaic panel through the onboard camera shooting module. The image information includes the condition of surface stains, damage or abnormality of the photovoltaic panel. Specifically, the image information acquisition module can be an imaging device such as a natural light camera or an infrared camera that can distinguish contours or heat sources; an image information processing module, which uses machine learning technology to analyze and identify the collected image information, so as to obtain the required operation instructions, identify different types of stains, estimate the severity of stains, and identify physical damage to the photovoltaic panel; specifically, using machine learning technology, connecting to the stain detection database, and comparing and analyzing the details of the image pixels, such as contours, colors, etc. The information on the picture or video is used to obtain the conditional instructions of the external environment I; the body monitoring module provides the photovoltaic cleaning robot with obstacle avoidance function through the body's own sensors and image information processing module, such as acoustic sensors and laser sensors, etc., to sense objects around the body and keep a certain distance from them, and monitor the body's power and the remaining amount of cleaning agent to obtain the conditional instructions inside the body II; the instruction judgment module is used to judge the priority of each conditional instruction and give execution instructions through the intelligent system, including a condition comparison submodule, which is used to judge the priority of conditional instructions; the instruction output submodule intelligently outputs the execution instruction according to the priority of the conditional instruction; the control processing module is used to control the photovoltaic cleaning robot to execute instructions.

[0073] The specific execution process of the intelligent cleaning mode is as follows:

[0074] S100, the control terminal issues a device startup command, and the photovoltaic cleaning robot enters the intelligent cleaning mode; the specific process is as follows:

[0075] S110, the control terminal issues a start command;

[0076] S120, determining the position information of the fuselage and returning an instruction to the terminal whether to inspect;

[0077] S121, when the control terminal chooses not to inspect, the stored cleaning route is retrieved according to the location information;

[0078] S122, when the control terminal selects inspection, the inspection range is manually set or the system defaults to the inspection range;

[0079] S123, when it is detected that the position change of the fuselage exceeds the preset range, the inspection range is manually set or the system defaults to the inspection range;

[0080] S130: If the intelligent system determines the inspection range, the body status and power level are checked and the water tank is drained.

[0081] Specifically, when the robot's positioning system detects that the change in the body position exceeds the preset range, the system will default to operating in the new environment, skip the control terminal's selection of whether to inspect, and directly start the forced inspection mode. The option subsequently fed back to the control terminal is the setting of the inspection range. The inspection range can be set on the control terminal, or it can be directly inspected according to the system's default mode, thereby improving the feasibility of the robot's intelligent operation; when the robot's positioning system detects that the body position change does not exceed the preset range, the system defaults to operating in the original environment. At this time, the system has stored the photovoltaic panel position information and the cleaning route map for the original environment. In this case, there will be two options, the control terminal chooses to inspect or not. When the control terminal chooses to inspect, the robot also enters the preparatory work of the inspection mode. When the control terminal chooses not to inspect, the intelligent control center calls the stored photovoltaic panel position information and the cleaning route map, and directly enters step S410 to start the cleaning operation, thereby improving the intelligent logic of the photovoltaic cleaning robot and improving the overall operation efficiency.

[0082] S200, establish a recognition model, automatically inspect based on the established recognition model, and collect photovoltaic panel location information; the specific process of establishing the recognition model is as follows:

[0083] S21, based on deep learning, uses lightweight GhostNet convolution calculation. The specific calculation formula is as follows:

[0084] ;

[0085] ;

[0086] in, express indivual aisle The convolution kernel size is Indicates A linear calculation is used to generate the Ghost feature map, represents the input feature map, represents the output feature map processed by the convolution layer, express Middle The channel feature map The linear operation results are: Indicates the number of channels, Represents an internal parameter, express Middle Channel feature maps;

[0087] S22. Use the CloU loss function to speed up and complete the model building. The specific expression is as follows:

[0088] ;

[0089] ;

[0090] ;

[0091] ;

[0092] ;

[0093] In the formula, Indicates that the center point of the predicted box and the real box is The distance of the direction, Indicates that the center point of the predicted box and the real box is The distance of the direction, represents the difference in bounding box lengths, represents the original loss function, represents the new improved loss function, and are the height and width of the feature map, respectively. is the regularization coefficient, is the standard deviation of the width and length of the prediction box. The specific process of automatic inspection based on the established recognition model and collecting photovoltaic panel location information is as follows:

[0094] S210, intelligent control center automatically plans inspection routes according to the inspection range;

[0095] S220, flying according to the inspection route, identifies the photovoltaic panels and stores the location information through machine learning technology, and maps it to the regional map.

[0096] Specifically, the photovoltaic cleaning robot will be equipped with a camera module to provide a precondition for machine learning technology comparison and identification. At the same time, during the inspection process, the intelligent control center will gradually form a large map of the inspection range in the system through the camera module. When a photovoltaic panel is identified, the location information of the photovoltaic panel will be mapped to the large map. This method uses artificial intelligence machine learning technology to ensure that the robot can identify photovoltaic panels by itself without human operation, while improving the efficiency of robot inspections.

[0097] S300, automatically planning a cleaning route and verifying learning based on the collected photovoltaic panel location information; the specific process is as follows:

[0098] S310, automatically planning a cleaning route according to the photovoltaic panel location information and performing an empty-load test flight;

[0099] S320, avoids obstacles and collects obstacle information through body sensors and machine learning technology;

[0100] S330. Replan the cleaning route according to the obstacle information and return the result to the control terminal.

[0101] Specifically, when the system has planned or manually stored a cleaning route, and this route has not yet been certified by the intelligent control center, the intelligent control center will control the photovoltaic cleaning robot to conduct an empty-load test flight according to the stored cleaning route to certify the route. If the route has been certified in the system records, the test flight certification link S300 will be skipped and the intelligent cleaning stage S400 will be directly entered. This improves the overall efficiency while ensuring the safe operation of the photovoltaic cleaning robot. In addition, the test flight stage is in an empty-load state, which is intended to verify the feasibility of the cleaning route, so that it can complete the verification quickly and ensure the lowest energy consumption.

[0102] S400, storing the cleaning route and controlling the photovoltaic cleaning robot to perform cleaning operations; the specific process is as follows:

[0103] S410, start intelligent cleaning, and the robot returns to the supply station to charge and add cleaning agent;

[0104] S420, based on the location information of the cleaning route, flying to and identifying the first photovoltaic panel of the route through the intelligent control center and machine learning technology;

[0105] S430, automatically identifying and judging whether the photovoltaic panel needs to be cleaned according to machine learning technology;

[0106] S431, automatically flying to the next photovoltaic panel on the cleaning route when cleaning is not needed, and recording the progress of the cleaning route;

[0107] S432, when cleaning is required, spray cleaning agent, judge the cleaning status of the photovoltaic panel, and after reaching the set cleaning standard, fly to the next photovoltaic panel on the route to continue the cleaning operation;

[0108] S440. When cleaning is completed, return to the supply station and record the progress of the cleaning route.

[0109] Specifically, when the photovoltaic cleaning robot enters the cleaning stage, it will first return to the supply station for replenishment. At this time, "Replenishing" will be displayed on the control terminal, and the robot's replenishment status will be displayed. After replenishment is completed, the intelligent control center will locate and identify the first photovoltaic panel according to the location information of the cleaning route, and start cleaning the photovoltaic panels in sequence. At the same time, during the cleaning process, machine learning technology is used to identify whether the current photovoltaic panel needs to be cleaned. If it is determined that cleaning is required, machine learning technology is used to identify the cleaning status of the photovoltaic panel during the cleaning process. After meeting the cleaning standards, it will fly to the next photovoltaic panel to continue the operation. When it is determined that cleaning is not required, it will fly directly to the next photovoltaic panel. When the cleaning operation is completed, the robot automatically returns to the supply station.

[0110] Preferably, implementing this mode through machine learning technology can effectively improve the operating efficiency of photovoltaic cleaning robots, while saving consumable resources such as cleaning agents, making the cleaning of photovoltaic panels more precise, especially in distributed photovoltaic power stations located in complex terrains such as mountains and roofs, which are particularly in need of such operating efficiency improvement and more intelligent and precise cleaning modes.

[0111] S500, according to different working stages, monitor the fuselage status and working status in real time during the operation process to determine whether to return to the supply station. The specific process is as follows:

[0112] S510, in the automatic inspection stage, the fuselage status and power are detected. When the fuselage status is abnormal or the power is insufficient, the aircraft returns to the supply station for replenishment. After replenishment, the aircraft automatically plans the shortest distance according to the progress of the operation route, and in combination with the obstacle avoidance system, flies directly from the supply station to the next operation block on the operation route;

[0113] S520, in the intelligent cleaning stage, the fuselage status, power and remaining cleaning agent are detected. When the fuselage status is abnormal or the power is insufficient, the robot returns to the supply station for replenishment. After replenishment, the robot automatically plans the shortest distance according to the progress of the operation route, and in combination with the obstacle avoidance system, flies directly from the supply station to the next section to be operated on the operation route;

[0114] It should be noted that the body's built-in sensors and machine learning technology are used to avoid obstacles throughout the entire stage. The obstacle avoidance system is turned on during the entire stage of the robot's flight. At the same time, the passing standards for the body status monitoring are carried out in stages. In addition, solutions are given to different flight requirements. For example, during the inspection stage, the photovoltaic cleaning robot turns off the monitoring of the remaining cleaning agent and turns on the monitoring of the robot's power. In the intelligent cleaning stage, the remaining cleaning agent and the robot's power are monitored at the same time. When one of the monitoring items does not meet the standard, the current work will be suspended, the system records the progress of the operation, and the robot automatically returns to the supply station for replenishment. When the replenishment is completed, the robot starts the operation again according to the stored operation progress until the cleaning operation is completed. The round-trip to the supply station during the operation process is carried out by raising the horizontal position and then returning in a straight line, which saves time. At the same time, the threshold of the monitoring item can be relatively lowered, thereby improving efficiency while ensuring the safe operation of the robot.

[0115] Therefore, the present invention adopts the above-mentioned intelligent system of the photovoltaic cleaning robot, and can achieve unmanned intelligent cleaning and operation and maintenance of photovoltaic panels even in distributed photovoltaic power stations in harsh environments and terrains. The operator only needs to operate the control terminal to issue simple instructions according to the attenuation of the power generation of the power station, and the photovoltaic cleaning robot can perform full-process intelligent operation and maintenance work, which greatly improves the convenience. In this application, the use of machine learning technology and the integration of obstacle avoidance systems greatly improve the rationality and safety of the logic of the intelligent operation and maintenance mode, and at the same time can improve the efficiency of photovoltaic intelligent cleaning operations.

[0116] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that they can still modify or replace the technical solution of the present invention with equivalents, and these modifications or equivalent replacements cannot cause the modified technical solution to deviate from the spirit and scope of the technical solution of the present invention.

Claims

1. An intelligent system for a photovoltaic cleaning robot, characterized in that: It includes a control terminal, an intelligent control center, a stain detection database and an intelligent device; the intelligent device includes a camera shooting module, a motion flight module and a cleaning module; the intelligent device and the intelligent control center are both integrated on the photovoltaic cleaning robot; the intelligent control center is connected to the control terminal through wireless communication; the intelligent control center is connected to the stain detection database through wireless communication; The intelligent control center receives a start command from the control terminal: when it is detected that the position of the fuselage exceeds the threshold of the last sleep position range, it returns to the control terminal to determine the inspection range; if there is no latitude change, it returns to the control terminal whether to inspect, and if it is selected not, the automatic cleaning mode is turned on; The intelligent control center is connected to the stain detection database in real time for continuous updating and learning of the machine learning algorithm; The intelligent control center is connected to the control terminal in real time to transmit the data and pictures of the cleaning operation or receive the instructions of the control terminal; The intelligent control center sends an execution instruction to the photovoltaic cleaning robot to control the photovoltaic cleaning robot to enter an intelligent cleaning mode; The specific execution process of the intelligent cleaning mode is as follows: S100, the control terminal issues a device start command, and the photovoltaic cleaning robot enters the intelligent cleaning mode; S200, establishing a recognition model, automatically inspecting based on the established recognition model, and collecting photovoltaic panel location information; S300, automatically planning a cleaning route and verifying learning based on the collected photovoltaic panel location information; S400, storing the cleaning route and controlling the photovoltaic cleaning robot to perform cleaning operations; S500, according to different working stages, real-time monitoring of the fuselage status and the working status during the operation process, and determining whether to return to the supply station; The specific process of establishing the recognition model in S200 is as follows: S21, based on deep learning, uses lightweight GhostNet convolution calculation. The specific calculation formula is as follows: , , in, express indivual aisle The convolution kernel size is Indicates A linear calculation is used to generate the Ghost feature map, represents the input feature map, represents the output feature map processed by the convolution layer, express Middle The channel feature map The linear operation results are: Indicates the number of channels, Represents an internal parameter, express Middle Channel feature maps; S22. Use the CloU loss function to speed up and complete the model building. The specific expression is as follows: In the formula, Indicates that the center point of the predicted box and the real box is The distance of the direction, Indicates that the center point of the predicted box and the real box is The distance of the direction, represents the difference in bounding box lengths, represents the original loss function, represents the new improved loss function, and are the height and width of the feature map, respectively. is the regularization coefficient, is the standard deviation of the width and length of the prediction box; In S100, the control terminal issues a device startup command, and the specific process of the photovoltaic cleaning robot entering the intelligent cleaning mode is as follows: S110, the control terminal issues a start command; S120, determining the position information of the fuselage and returning an instruction to the terminal whether to inspect; S130, if the intelligent system determines the inspection range, then the machine body status is detected, the power level is detected, and the water tank is emptied; The instructions for determining the location information of the fuselage and returning the terminal to whether to inspect in S120 specifically include: S121, when the control terminal chooses not to inspect, the stored cleaning route is retrieved according to the location information; S122, when the control terminal selects inspection, the inspection range is manually set or the system defaults to the inspection range; S123, when it is detected that the position change of the fuselage exceeds the preset range, the inspection range is manually set or the system defaults to the inspection range; The specific process of automatic inspection based on the established recognition model in S200 and collecting photovoltaic panel location information is as follows: S210, intelligent control center automatically plans inspection routes according to the inspection range; S220, flying according to the inspection route, identifying the photovoltaic panels and storing the location information through machine learning technology, and mapping it to the regional map; The specific process of automatically planning the cleaning route and verifying the learning based on the collected photovoltaic panel location information in S300 is as follows: S310, automatically planning a cleaning route according to the photovoltaic panel location information and performing an empty-load test flight; S320, avoids obstacles and collects obstacle information through body sensors and machine learning technology; S330, replanning the cleaning route according to the obstacle information and returning the result to the control terminal; The specific process of storing the cleaning route in S400 to control the photovoltaic cleaning robot to perform cleaning operations is as follows: S410, start intelligent cleaning, and the robot returns to the supply station to charge and add cleaning agent; S420, based on the location information of the cleaning route, flying to and identifying the first photovoltaic panel of the route through the intelligent control center and machine learning technology; S430, automatically identifying and judging whether the photovoltaic panel needs to be cleaned according to machine learning technology; S431, automatically flying to the next photovoltaic panel on the cleaning route when cleaning is not needed, and recording the progress of the cleaning route; S432, when cleaning is required, spray cleaning agent, judge the cleaning status of the photovoltaic panel, and after reaching the set cleaning standard, fly to the next photovoltaic panel on the route to continue the cleaning operation; S440, when the cleaning is completed, return to the supply station and record the progress of the cleaning route; The specific process of determining whether to return to the supply station in S500 is as follows: S510, in the automatic inspection stage, the fuselage status and power are detected. When the fuselage status is abnormal or the power is insufficient, the aircraft returns to the supply station for replenishment. After replenishment, the aircraft automatically plans the shortest distance according to the progress of the operation route, and in combination with the obstacle avoidance system, flies directly from the supply station to the next operation block on the operation route; S520: During the intelligent cleaning stage, the fuselage status, power level and remaining cleaning agent are detected. When an abnormal fuselage status or insufficient power is detected, the robot returns to the supply station for replenishment. After replenishment, the robot automatically plans the shortest distance based on the progress of the operation route, and in combination with the obstacle avoidance system, flies directly from the supply station to the next section to be operated on the operation route.

2. The intelligent system of a photovoltaic cleaning robot according to claim 1, characterized in that: The smart device also includes: The image acquisition module collects the image information of the photovoltaic panel through the onboard camera shooting module. The image information includes the surface stains, damage or abnormality of the photovoltaic panel; The image information processing module uses machine learning technology to analyze and identify the collected image information, obtain the required operating instructions, identify different types of stains, estimate the severity of stains, and identify physical damage to photovoltaic panels; The body monitoring module provides obstacle avoidance for the photovoltaic cleaning robot through the body's built-in sensors combined with the image information processing module, monitors the body's power and the remaining amount of cleaning agent, and obtains conditional instructions II; The instruction judgment module is used to judge the priority of each conditional instruction and give the execution instruction through the intelligent system, including: a condition comparison submodule, which is used to judge the priority of the conditional instruction; an instruction output submodule, which intelligently outputs the execution instruction according to the priority of the conditional instruction; The control processing module is used to control the photovoltaic cleaning robot to execute instructions.

Citation Information

Cited By

  • Photovoltaic steam self-cleaning system based on multi-source sensing and waste heat cooperation and control method

    CN122026792A