Photovoltaic precise cleaning method and system based on cooperation of two unmanned aerial vehicles

Abstract

The application discloses a photovoltaic precise cleaning method and system based on cooperation of double unmanned aerial vehicles, and belongs to the technical field of photovoltaic cleaning. A first unmanned aerial vehicle inspects a target area of a photovoltaic module and collects image data, a ground station identifies and classifies pollution according to the image data, and determines a cleaning mode according to a pollution classification result. In the precise cleaning mode, the first unmanned aerial vehicle sprays a cleaning agent to a polluted area, and after a preset action time, a second unmanned aerial vehicle performs water flushing to the area that has been sprayed with the cleaning agent; after cleaning is completed, the first unmanned aerial vehicle collects re-inspection image data again, and the ground station judges whether the cleaned area reaches a preset cleaning requirement, and re-executes cleaning and re-inspection for the area that does not reach the requirement. The application forms a continuous closed-loop cleaning process through inspection, task decision, double-machine cooperative cleaning and re-inspection feedback, and improves the pertinence and integrity of photovoltaic module cleaning operation.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicles (UAVs), specifically to a method and system for precise photovoltaic cleaning based on dual UAV collaboration. Background Technology

[0002] As the scale of photovoltaic power plant construction continues to expand, photovoltaic modules are exposed to the outdoor environment for extended periods, making their surfaces prone to accumulating dust, bird droppings, oil, and other pollutants. These pollutants can block the sunlight-receiving area of ​​the photovoltaic modules, affecting their normal power generation. Therefore, regular inspection and cleaning of photovoltaic modules are necessary. Current photovoltaic module cleaning methods are gradually shifting from manual cleaning to drone cleaning. Drones equipped with cameras acquire images of the photovoltaic module surface or use spray systems to clean the modules, reducing the workload of manual entry into large-area photovoltaic arrays. Existing technologies also include solutions that utilize different equipment to perform inspection and cleaning operations separately.

[0003] However, in actual cleaning processes, there is often a lack of continuous task correlation between the pollution information obtained from inspections and subsequent cleaning operations. This can easily lead to situations where cleaning is carried out in the same manner after inspections, making it difficult to determine the appropriate cleaning mode based on the identified pollution levels. Furthermore, when a single drone is used to sequentially handle inspection and cleaning tasks, multiple operational stages are performed serially, which is not conducive to continuously completing cleaning agent treatment and water rinsing of identified contaminated areas. This makes it difficult to promptly translate inspection results into cleaning tasks tailored to specific contaminated areas. Especially in cases of stubborn localized pollution, simply using direct water rinsing or cleaning the entire area in the same way fails to balance targeted treatment of the target area with overall operational efficiency.

[0004] Furthermore, some existing cleaning methods end the operation once cleaning is complete, lacking an automatic feedback relationship between the cleaning results and previous cleaning tasks. When pollutants remain in a local area, it is necessary to rely on personnel to re-inspect and reorganize the cleaning process. Therefore, how to use the pollution information obtained from inspections to determine the appropriate cleaning mode, use different drones to collaboratively complete the cleaning agent treatment and water rinsing of the contaminated area, and use the re-inspection results after cleaning to determine whether further cleaning is needed, thus forming a continuous closed-loop operation process from pollution identification, cleaning execution to cleaning effect verification, is a specific technical problem that needs to be solved in the unmanned cleaning of photovoltaic modules. Summary of the Invention

[0005] The purpose of this invention is to provide a photovoltaic precision cleaning method and system based on dual-UAV collaboration. By forming a continuous closed-loop cleaning process through inspection, task decision-making, dual-UAV collaborative cleaning, and re-inspection feedback, the cleaning operation of photovoltaic modules is improved in terms of targeting and completeness.

[0006] To achieve the above objectives, the present invention proposes the following technical solution: a photovoltaic precision cleaning method based on dual-UAV collaboration, comprising the following steps:

[0007] S1: The first UAV performs an inspection of the target area of ​​the photovoltaic module and collects image data of the target area;

[0008] S2: The ground station receives the image data and performs pollution identification and classification based on the image data to obtain pollution classification results;

[0009] S3: The ground station determines the cleaning mode based on the pollution classification results and generates a corresponding cleaning task based on the determined cleaning mode.

[0010] S4: When the cleaning mode is the precision cleaning mode, the first drone flies to the identified contaminated area and sprays cleaning agent on the contaminated area. The first drone only sprays the identified contaminated area at a fixed point and does not spray non-contaminated areas.

[0011] S5: After a preset action time, the second drone flies to the contaminated area that has been sprayed with the cleaning agent and performs a water rinse on the contaminated area. The preset action time is dynamically set within the range of 30 seconds to 3 minutes depending on the type of contamination.

[0012] S6: After rinsing with clean water, the first UAV will perform a re-inspection of the cleaned contaminated area and collect re-inspection image data;

[0013] S7: The ground station determines whether the cleaned contaminated area meets the preset cleaning requirements based on the re-inspection image data. The preset cleaning requirements are that the residual stain area is less than 2cm×2cm. When it is determined that the preset cleaning requirements are not met, the ground station controls the first UAV and the second UAV to re-clean and re-inspect the contaminated area that does not meet the preset cleaning requirements.

[0014] Furthermore, in this invention, in step S1, the first drone collects image data of the photovoltaic module using a visible light camera and an infrared thermal imaging camera; in step S2, the location, type, and area of ​​pollutants are determined based on the image data, and the pollutants are classified into Class A pollution, Class B pollution, and Class C pollution, wherein Class A pollution is stubborn stains, Class B pollution is moderate pollution, and Class C pollution is light pollution; Class A pollution includes at least one of bird droppings, oil stains, and cement spots, Class B pollution includes regional dust accumulation, and Class C pollution includes large areas of floating dust.

[0015] Furthermore, in this invention, in step S3, the ground station calculates the area proportion of the Class A contaminated area in all contaminated areas and the area proportion of the Class C contaminated area in all contaminated areas; when the area proportion of the Class A contaminated area in all contaminated areas is greater than 10%, the precision cleaning mode is automatically activated, in which only the Class A contaminated areas are pretreated with cleaning agents and rinsed at specific points; when the area proportion of the Class C contaminated area in all contaminated areas is greater than 80%, the indiscriminate cleaning mode is automatically activated, in which large-area clean water spraying is performed on all target areas; in other cases, the operator selects the cleaning mode according to the contamination distribution map.

[0016] Furthermore, in the present invention, in the precise cleaning mode, the ground station plans the coordinated flight path of the first UAV and the second UAV according to the distribution of the polluted area, so that the flight path of the first UAV preferentially passes through the Class A polluted area, and the time when the second UAV arrives at the corresponding polluted area has a time interval relative to the time when the first UAV sprays the cleaning agent, which is the preset action time; the preset action time is 30 seconds to 3 minutes, and the time interval is dynamically adjusted according to the type of pollution.

[0017] Furthermore, in this invention, the first drone uses a foam nozzle to spray the cleaning agent. The foam nozzle sprays the cleaning agent in foam form through an atomizing core assembly, and the spray range and coverage width of the foam are limited by a conical diffuser set at the nozzle. The second drone selects a clean water nozzle with different spray angles according to the type of pollution. For stubborn stains, a 0-degree direct spray nozzle or a 15-degree narrow spray nozzle is used; for large areas of light pollution, a 40-degree wide spray nozzle is used; and for areas of dust accumulation, a 25-degree medium spray nozzle or a 40-degree wide spray nozzle is used.

[0018] Furthermore, in this invention, during the operation of the first drone and / or the second drone, the battery level and liquid level of the corresponding drone are monitored in real time. When the battery level or liquid level is lower than a preset threshold, the corresponding drone is controlled to return to base for battery replenishment and / or liquid replenishment, and the unfinished cleaning task is continued after replenishment is completed. In step S7, whether the area of ​​residual stains in the cleaning area is less than the preset residual stain area is used as the basis for judging whether the cleaning meets the standard.

[0019] Furthermore, in this invention, when the indiscriminate cleaning mode is activated, both the first UAV and the second UAV carry clean water nozzles. The ground station plans a large-area coverage route for the two UAVs in coordination based on the distribution of the photovoltaic modules to be cleaned. This allows the first UAV and the second UAV to stagger their work areas along the coordinated large-area coverage route and jointly perform large-area clean water spray cleaning of the photovoltaic modules.

[0020] A photovoltaic precision cleaning system based on dual-UAV collaboration is used to implement the aforementioned photovoltaic precision cleaning method based on dual-UAV collaboration. The system is characterized by comprising a first UAV, a second UAV, and a ground station.

[0021] The first UAV is equipped with an inspection payload and a replaceable cleaning payload, used to collect inspection image data of photovoltaic modules, spray cleaning agent on the polluted area according to the cleaning task issued by the ground station, and collect re-inspection image data after cleaning is completed.

[0022] The second UAV is equipped with a water spraying assembly, which is used to rinse the contaminated area that has been sprayed with cleaning agent with water according to the cleaning task issued by the ground station;

[0023] The ground station is communicatively connected to the first UAV and the second UAV respectively, and is equipped with a cleaning task scheduling module. The cleaning task scheduling module is used to identify and classify pollution based on the inspection image data, determine the cleaning mode based on the pollution classification result and schedule the first UAV and the second UAV to perform the corresponding cleaning task, and determine whether the cleaning meets the standard based on the re-inspection image data, and trigger a supplementary cleaning task when it is determined that the standard is not met.

[0024] Furthermore, in this invention, the inspection payload of the first UAV includes a visible light camera and an infrared thermal imaging camera. The visible light camera has 2 megapixels, and the infrared thermal imaging camera has a resolution of 640×512. The replaceable cleaning payload of the first UAV includes a cleaning agent foam nozzle. The cleaning agent foam nozzle includes an atomizing core assembly and a conical diffuser shroud disposed at the nozzle, and is connected to the first UAV via a quick-release interface.

[0025] The second UAV's water spray assembly includes a water tank, a high-pressure pump, a water nozzle assembly, and a two-degree-of-freedom gimbal. The water tank has a capacity of 30 to 50 liters and is equipped with an anti-sway structure. The high-pressure pump has an adjustable water pressure range of 0 to 12 MPa and a rated flow rate of not less than 10 L / min. The water nozzle assembly includes a 0-degree direct spray nozzle, a 15-degree narrow spray nozzle, a 25-degree medium spray nozzle, and a 40-degree wide spray nozzle. Each of the water nozzles is installed via a quick-release interface. The two-degree-of-freedom gimbal has a pitch range of -60 degrees to +35 degrees and a horizontal range of ±45 degrees.

[0026] Furthermore, in this invention, the cleaning task scheduling module is configured as follows:

[0027] Receive inspection image data collected by the first UAV, and obtain the location, type and area of ​​pollutants based on the inspection image data;

[0028] Pollution is classified into three categories: Category A, Category B, and Category C. Category A pollution represents stubborn stains, Category B pollution represents moderate pollution, and Category C pollution represents light pollution.

[0029] Calculate the area percentage of the Class A and Class C contaminated areas in all contaminated areas, and determine whether to activate the precision cleaning mode or the indiscriminate cleaning mode based on the area percentage.

[0030] In the precision cleaning mode, a dual-drone collaborative route is planned, so that the first drone arrives at the contaminated area first to spray cleaning agent, and the second drone arrives at the corresponding contaminated area after a preset action time to perform water rinsing.

[0031] In the indiscriminate cleaning mode, a dual-drone collaborative large-area coverage route is planned so that the first drone and the second drone work in staggered areas to jointly perform clean water spray cleaning.

[0032] It also receives re-inspection image data collected by the first drone, determines whether the cleaning area meets the preset cleaning requirements based on the re-inspection image data, and generates a supplementary cleaning task for the unqualified contaminated area if the preset cleaning requirements are not met.

[0033] Beneficial effects: The technical solution of this application has the following technical effects:

[0034] This invention first utilizes a first UAV to inspect the target area of ​​photovoltaic modules and obtain image data. A ground station then identifies and classifies contamination based on this image data, and further determines the corresponding cleaning mode based on the contamination classification results. This allows subsequent cleaning tasks to be built upon the results of the preceding inspection. Therefore, the inspection operation is no longer merely used to obtain the surface condition of the photovoltaic modules, but can also serve as a basis for selecting the cleaning mode and generating cleaning tasks. This facilitates targeted cleaning of already identified contaminated areas and reduces the problem of discrepancies between inspection information and actual cleaning execution.

[0035] In precision cleaning mode, a first drone sprays cleaning agent onto the contaminated area, followed by a second drone rinsing the area with clean water after a preset processing time. This allows the cleaning agent treatment and rinsing to be completed collaboratively by the two drones in a sequential manner. Compared to a single drone handling all cleaning steps sequentially, this dual-drone approach assigns different cleaning tasks to corresponding drones and creates a continuous process between cleaning agent pretreatment and subsequent water rinsing. This improves the continuity of cleaning contaminated areas and enhances the cleaning effect on attached contaminants.

[0036] After rinsing with clean water, the first drone re-inspects the cleaned area, and the ground station determines whether the cleaned area meets the preset cleaning requirements based on the re-inspection image data. For areas that do not meet the cleaning requirements, the first and second drones are re-controlled to perform the corresponding cleaning and re-inspection processes. This forms a closed-loop operation process of "inspection identification - pattern determination - collaborative cleaning - re-inspection judgment - cleaning again if necessary," allowing areas with residual contamination to continue to enter the subsequent cleaning process. This reduces the need for complete manual inspection and reorganization after cleaning, and improves the integrity of the photovoltaic module cleaning process and the reliability of the cleaning results. Attached Figure Description

[0037] Figure 1 This is a structural block diagram of the photovoltaic precision cleaning system based on dual-drone collaboration of the present invention (in the figure, the multi-functional drone and the cleaning drone correspond to the first drone and the second drone, respectively).

[0038] Figure 2 This is a flowchart of the precise cleaning mode method of the present invention. Detailed Implementation

[0039] The technical solution of the present invention will be further described in detail below with reference to specific embodiments. It should be noted that the following embodiments are used to illustrate the technical principles and specific implementation methods of the present invention, and are not intended to limit the scope of protection of the present invention. The parameters such as equipment capacity, identification accuracy, spray pressure, time threshold, area ratio, and cleaning judgment criteria involved in the embodiments can be adjusted accordingly based on the scale of the photovoltaic power station, the arrangement of photovoltaic modules, the degree of pollution, and the on-site environment.

[0040] Example 1

[0041] This embodiment provides a photovoltaic precision cleaning system based on dual-UAV collaboration, including a first UAV, a second UAV, and a ground station. The first and second UAVs establish communication connections with the ground station, which receives inspection and re-inspection image data collected by the first UAV. Based on the image data, the ground station performs contamination identification, contamination classification, cleaning mode selection, and cleaning task scheduling, while simultaneously sending corresponding flight and cleaning tasks to both UAVs. The first UAV primarily undertakes photovoltaic module inspection, cleaning agent spraying, and post-cleaning re-inspection tasks, while the second UAV primarily undertakes water rinsing. This allows the two UAVs to form a coordinated cleaning operation under the unified scheduling of the ground station.

[0042] The first UAV is equipped with an inspection payload and a replaceable cleaning payload. The inspection payload includes a visible light camera and an infrared thermal imaging camera. The visible light camera can be a 2-megapixel camera, and the infrared thermal imaging camera has a resolution of 640×512. Using these inspection payloads, the first UAV can identify contaminants with a minimum size of 2cm×2cm and hot spots with a minimum size of 3cm×3cm, and can obtain information on the location, type, and area of ​​contaminants on the photovoltaic module surface.

[0043] The first type of anomaly that can be identified during UAV inspection includes diode failures, internal defects, string short circuits or open circuits, object obstruction, surface stains, component damage, component misalignment, and edge dust accumulation. Internal defects can include microcracks, hot spots, and delamination. For the cleaning task of this invention, the focus is on using information such as surface stains, object obstruction, and edge dust accumulation to determine the contamination status of the photovoltaic module surface. Simultaneously, infrared images can also help determine whether there are abnormal temperature rise areas in the photovoltaic module caused by localized obstruction or contamination.

[0044] The replaceable cleaning payload of the first UAV includes a cleaning agent foam nozzle. The cleaning agent foam nozzle contains an atomizing core assembly, and a conical diffuser is positioned at the nozzle outlet. The atomizing core assembly atomizes the cleaning agent into foam before spraying it, allowing the foam to cover the surface of the target contaminants. The conical diffuser limits and adjusts the diffusion range and coverage width of the foam, facilitating the first UAV to spray cleaning agent onto identified localized contaminated areas. The cleaning agent foam nozzle connects to the first UAV via a unified quick-release interface, allowing for nozzle replacement within approximately 3 seconds during implementation. This enables the first UAV to switch payloads between inspection and cleaning agent spraying tasks.

[0045] The second UAV is equipped with a water spray system, which includes a water tank, a high-pressure pump, a water nozzle assembly, and a two-degree-of-freedom gimbal. The water tank capacity can be set to 30L to 50L, and an anti-sway structure is installed inside the tank. The anti-sway structure is used to reduce the impact of large water swaying on the UAV's center of gravity and flight attitude during acceleration, deceleration, turning, or hovering adjustments, thereby improving the stability of the UAV during water-carrying flight.

[0046] The high-pressure pump is used to deliver clean water from the water tank to the clean water nozzle assembly. The water pressure of the high-pressure pump is adjustable from 0 to 12 MPa, and the rated flow rate is not less than 10 L / min. By adjusting the water supply pressure and combining it with different nozzles, the impact of the water flow and the cleaning coverage can be adjusted according to the degree of contaminant adhesion, thereby meeting the needs of different cleaning scenarios such as localized stubborn contamination and large-area light contamination.

[0047] The clean water spray head assembly includes a 0-degree direct spray head, a 15-degree narrow spray head, a 25-degree medium spray head, and a 40-degree wide spray head. All spray heads can be installed using quick-release interfaces. The 0-degree direct spray head creates a relatively concentrated water flow, suitable for stubborn stains such as bird droppings; the 15-degree narrow spray head provides both coverage and concentrated rinsing, suitable for precise rinsing of small contaminated areas; the 25-degree medium spray head is suitable for general contamination such as accumulated dust; and the 40-degree wide spray head offers a wider cleaning coverage, suitable for light contamination such as large areas of floating dust.

[0048] The two-degree-of-freedom gimbal is used to adjust the spray direction of the water nozzle. Its pitch adjustment range is -60 degrees to +35 degrees, and its horizontal adjustment range is ±45 degrees. After the second UAV flies to the corresponding polluted area, it can adjust the nozzle direction through the two-degree-of-freedom gimbal while maintaining a basic flight attitude, so that the water can be accurately applied to the target area on the surface of the photovoltaic module.

[0049] The ground station is equipped with a cleaning task scheduling module. This module receives visible light and infrared images from the first UAV, identifies the location, type, and area of ​​contamination on the photovoltaic module surface, and classifies the contamination into three categories: Category A, Category B, and Category C. Category A contamination is persistent, including at least one of bird droppings, oil stains, and cement spots; Category B contamination is moderate, such as accumulated dust; and Category C contamination is light, such as large areas of floating dust.

[0050] After completing the pollution classification, the ground station further calculates the area of ​​each type of polluted area. Specifically, the ground station calculates the proportion of Class A polluted areas and the proportion of Class C polluted areas. When the proportion of Class A polluted areas is greater than 10%, it indicates that there is a certain amount of stubborn pollution in the target area that requires pre-softening treatment, so the precision cleaning mode is activated; when the proportion of Class C polluted areas is greater than 80%, it indicates that the target area is mainly a large area of ​​light pollution, so the indiscriminate cleaning mode is activated; when none of the above conditions are met, the operator can manually select the appropriate cleaning mode based on the actual pollution distribution.

[0051] When the precision cleaning mode is activated, the ground station first plans the coordinated flight path of the first and second UAVs based on the pollution distribution map, prioritizing the flight path of the first UAV over Class A pollution areas. After the first UAV replaces the foam nozzle of the cleaning agent, it flies to the corresponding pollution area according to the planned route and sprays foam cleaning agent onto the target pollutants while hovering or flying at low speed.

[0052] After the cleaning agent spraying is completed, the cleaning agent is allowed to remain on the contaminant surface for a preset time. In this embodiment, the preset time is set to 30 seconds to 3 minutes, and can be configured according to the degree of contamination, the type of cleaning agent, and the on-site temperature. For example, the time can be set to about 1 minute to allow the cleaning agent to fully wet and soften bird droppings and other attached contaminants before the second drone performs a water rinse.

[0053] After a preset treatment time, the second UAV flies along the coordinated flight path to the area already sprayed with cleaning agent. For stubborn stains such as bird droppings, a 0-degree direct spray nozzle or a 15-degree narrow spray nozzle can be used for rinsing; for accumulated dust in a region, a 25-degree medium spray nozzle can be used; and for large areas of floating dust, a 40-degree wide spray nozzle can be used. The second UAV delivers clean water to the corresponding nozzles via a high-pressure pump and adjusts the spray direction using a two-degree-of-freedom gimbal, thereby completing the clean water rinsing of the corresponding contaminated area.

[0054] During the flight and cleaning process of the first and second drones, the system monitors the remaining battery power and corresponding liquid levels of both drones in real time. When the drone's battery power falls below a preset threshold, the system controls the corresponding drone to return to base to replenish its battery; when the remaining liquid level of cleaning agent or water falls below a preset threshold, the system controls the corresponding drone to return to base to replenish the liquid. After replenishment is complete, the corresponding drone re-enters the area where the task was not completed to continue the cleaning mission.

[0055] For the second drone, the remaining water volume can be determined by the liquid flow status. One parameter setting method in this embodiment is: when the difference between the inlet and outlet flow rates is less than 1L, the water tank is considered nearly empty, triggering a return-to-base resupply. It should be noted that this parameter is an exemplary parameter in a specific implementation and can be adjusted according to the water tank capacity, flow detection structure, and actual cleaning task.

[0056] After all contaminated areas have been rinsed with clean water, the first UAV is reloaded with its inspection payload or resumes inspection mode and flies back to the cleaned area to collect re-inspection images. Once the re-inspection images are returned to the ground station, the ground station identifies any remaining dirt after cleaning and determines whether the cleaning quality meets the preset requirements.

[0057] In this embodiment, a residual stain area of ​​less than 2cm × 2cm can be used as the cleaning standard. When a re-inspection confirms that the residual stain in the target area meets this standard, the corresponding area is marked as cleaned. When the residual stain in a certain location still does not meet the standard, the ground station re-identifies that location as the area to be cleaned, and re-schedules the first UAV to spray cleaning agent and the second UAV to perform water rinsing. Then, the first UAV is used again for re-inspection, thus forming a closed-loop process of cleaning-re-inspection-supplementary cleaning.

[0058] Example 2

[0059] This embodiment uses the precise cleaning operation of a tidal flat photovoltaic power station as an example. The photovoltaic array is located near a floating body on the water surface or in a tidal flat environment, and is easily affected by bird activity and sea breeze dust during long-term operation. The surface of the photovoltaic modules is simultaneously contaminated with bird droppings and accumulated dust. In the operation area corresponding to this embodiment, there are an average of about 3 to 5 bird droppings contaminations per row of photovoltaic modules.

[0060] After the operation begins, the first UAV, equipped with a 2-megapixel visible light camera and a 640×512 resolution infrared thermal imaging camera, performs a patrol inspection along the photovoltaic array and transmits the collected images to the ground station in real time or in batches. The ground station uses the images to determine the location and area of ​​polluted areas such as bird droppings and accumulated dust, and classifies them into Class A, Class B, and Class C pollution categories.

[0061] According to ground station statistics, in this embodiment, Class A pollution, i.e., bird droppings pollution areas, account for approximately 15% of the total polluted area; Class C pollution, i.e., large areas of accumulated ash or dust, accounts for approximately 70% of the total polluted area. Since the Class A pollution area accounts for approximately 15%, which is greater than the 10% threshold for triggering the precision cleaning mode, the ground station automatically activates the precision cleaning mode.

[0062] The ground station plans the coordinated flight path of the first and second UAVs based on the location of bird droppings contamination, prioritizing the first UAV's passage over all Class A contamination areas. The first UAV uses a quick-release interface to replace the foam cleaning agent nozzle, a process that can be completed in approximately 3 seconds. Subsequently, the first UAV flies sequentially to the bird droppings contamination locations along the planned route and sprays foam cleaning agent at the corresponding locations.

[0063] In this embodiment, the action time of the cleaning agent on the bird droppings contaminant surface is set to approximately 1 minute, which is within a configurable range of 30 seconds to 3 minutes. After spraying, the foam cleaning agent adheres to the bird droppings surface and, during the approximately 1-minute waiting period, wets and softens the droppings.

[0064] After about one minute, the second drone flew to the corresponding bird droppings contaminated area to rinse it with clean water. For stubborn stains such as bird droppings, a 0-degree direct spray nozzle or a 15-degree narrow spray nozzle can be used for precise rinsing; for large areas of accumulated dust, a 25-degree medium spray nozzle or a 40-degree wide spray nozzle can be selected to increase the cleaning area covered per unit flight distance.

[0065] The second UAV is equipped with a 30L-50L water tank, with an anti-sway structure inside. The high-pressure pump has a pressure adjustment range of 0-12MPa and a rated flow rate of not less than 10L / min. Its two-degree-of-freedom gimbal has a pitch adjustment range of -60 degrees to +35 degrees and a horizontal adjustment range of ±45 degrees. During the flushing process, the nozzle attitude and flushing direction can be adjusted according to the location of the target contaminated area.

[0066] During this actual operation, the second drone was triggered to return to base for water replenishment when its water tank level was close to empty. In one specific method of judgment, the difference between the inlet and outlet flow rates being less than 1L can be used as a parameter to determine if the tank is nearing empty. In this embodiment, two replenishments were performed during the entire cleaning process. After each replenishment, the second drone returned to the unfinished area to continue cleaning.

[0067] After all contaminated areas were rinsed, the first drone took off again to re-inspect the cleaning quality. The ground station determined the area of ​​residual stains based on the re-inspection images, and a residual stain area of ​​less than 2cm × 2cm was used as the cleaning standard for this embodiment. The re-inspection results showed that all contaminated areas in this embodiment met this cleaning standard, therefore, no further cleaning was required.

[0068] Under the above operating conditions, the total time to complete the entire cleaning task in this embodiment is approximately 35 minutes, including two return resupply processes; the entire cleaning task consumes approximately 65L of clean water and approximately 3L of cleaning agent. For stubborn stains such as bird droppings, pretreatment with foam cleaning agent followed by rinsing with clean water achieves a stain removal rate of over 98%; in comparison, rinsing directly with pure clean water results in a removal rate of approximately 70% for stubborn stains such as bird droppings.

[0069] Under the operating conditions of this embodiment, the first drone is responsible for inspection and cleaning agent pretreatment, while the second drone handles water rinsing. This allows different tasks to be executed along a coordinated route, increasing cleaning efficiency by approximately 100% compared to the method where a single drone performs the corresponding tasks sequentially. Simultaneously, since the cleaning agent and high-intensity rinsing are primarily applied to identified contaminated areas rather than spraying all photovoltaic modules to the same degree, water consumption is reduced by approximately 40% compared to indiscriminate cleaning. The above performance data corresponds to the specific operating scenario and parameter conditions of this embodiment and are used to illustrate the actual operating results of this implementation method, but do not constitute a limitation on the scope of protection of this invention.

[0070] Example 3

[0071] This embodiment illustrates the indiscriminate cleaning process in a large-area, lightly polluted scenario. For example, after the first centralized cleaning of a photovoltaic power station or after a long operating cycle, the surface of its components mainly forms a large area of ​​floating dust, without a large number of stubborn stains such as bird droppings or oil stains.

[0072] The first UAV uses its inspection payload to collect images of the target area and transmits the image data to the ground station. After pollution identification and area statistics, in this embodiment, the Class C lightly polluted area accounts for approximately 90% of the total polluted area, exceeding the 80% threshold for triggering the indiscriminate cleaning mode. Therefore, the ground station automatically activates the indiscriminate cleaning mode.

[0073] In the indiscriminate cleaning mode, both the first and second drones are equipped with clean water nozzles. Both drones can use 40-degree wide-angle nozzles to achieve a larger single-spray coverage area. The ground station plans the dual-drone large-area coverage routes based on the spatial distribution of the entire photovoltaic array, and staggers the operating areas of the two drones to avoid flight path conflicts in the same area.

[0074] The first and second drones work together to perform water spraying cleaning according to their respective assigned coverage areas. During flight, the drones' battery level and liquid level are continuously monitored in real time. When a drone's battery or liquid level is insufficient, it is controlled to return to base for resupply. After resupply, it returns to the unfinished cleaning area to continue working until the entire planned area is covered. Because both drones are responsible for water spraying in this mode, they can cover different areas simultaneously, making it suitable for lightly polluted scenarios with large areas dominated by floating dust.

[0075] Example 4

[0076] This embodiment further illustrates the mixed pollution cleaning method when different types of pollution coexist. In an actual photovoltaic array, bird droppings, edge dust accumulation, and large areas of floating dust may coexist. Therefore, a first UAV can be used to acquire visible light and infrared images of the entire target area, and then a ground station can comprehensively determine the area proportion of each type of pollution.

[0077] In this embodiment, according to the ground station's identification and statistics, Class A pollution area accounts for approximately 12% of the total polluted area, and Class C pollution area accounts for approximately 60% of the total polluted area. Since the Class A pollution area accounts for approximately 12%, which exceeds the 10% trigger threshold for the precision cleaning mode, and the Class C pollution area accounts for approximately 60%, which is lower than the 80% trigger threshold for the indiscriminate cleaning mode, the ground station activates the precision cleaning mode.

[0078] The ground station first plans the priority flight path for the first drone based on the Class A pollution location. The first drone, carrying a foam nozzle, arrives at the Class A pollution area, such as bird droppings, and sprays foam cleaning agent onto the corresponding pollution area. After the set cleaning agent action time of 30 seconds to 3 minutes, the second drone arrives at the corresponding area and uses a 15-degree narrow-width nozzle to precisely wash the Class A pollution location.

[0079] For the remaining Class B areas with accumulated dust and Class C areas with large areas of floating dust, the second drone switches to either a 25-degree medium-width nozzle or a 40-degree wide-width nozzle to perform clean water rinsing, depending on the type of pollution. Thus, in the same photovoltaic array cleaning mission, stubborn contamination is treated with cleaning agent pretreatment and narrow-width concentrated rinsing, while moderate or light contamination is treated with nozzles covering a larger area, ensuring that different contaminated areas are cleaned using methods appropriate to their contamination levels.

[0080] After all areas have been cleaned, the first drone will conduct a second inspection of the cleaned areas. If the area of ​​residual stains is less than 2cm × 2cm, the cleaning is considered to have met the standard. If there are still areas that do not meet the standard, the ground station will generate a new cleaning task based on the location of the residual contaminants, and continue to dispatch the first and second drones to complete the spraying of cleaning agent, rinsing with clean water, and a second inspection until the preset cleaning requirements are met.

[0081] Example 5

[0082] This embodiment, based on the above embodiments, further provides a dual-UAV collaborative cleaning method based on the time window of the contamination task point. It is mainly applicable to situations where multiple dispersed, stubborn contamination areas exist simultaneously on the surface of photovoltaic modules. The remaining UAV structure, cleaning nozzles, ground station, and contamination identification method can adopt the structure and method described in the above embodiments. This embodiment focuses on explaining how the second UAV completes water rinsing according to the actual action time of different contamination areas after the first UAV completes the cleaning agent spraying.

[0083] After the first UAV completes its inspection of the photovoltaic modules, the ground station identifies multiple contaminated areas to be treated based on the inspection images, designating each area as an independent contamination task point. Each contamination task point corresponds to at least the location of the contaminated area and the type of contamination. The first UAV arrives at each contamination task point sequentially according to the flight route planned by the ground station and sprays cleaning agent onto the corresponding contaminated area through its foam nozzles.

[0084] Unlike the method of uniformly timing all contaminated areas, in this embodiment, the ground station records the completion time of the cleaning agent spraying at each contaminated task point after the first UAV completes the spraying. Therefore, even if the first UAV continuously processes multiple contaminated task points, each contaminated task point has its own starting time for the cleaning agent action.

[0085] The ground station determines the permissible rinsing time window for each contaminated task point based on the type of contamination and the corresponding spray completion time. This permissible rinsing time window falls within the range of 30 seconds to 3 minutes of cleaning agent action time. For different contaminated task points, different action times can be used depending on the degree of contaminant adhesion, thus allowing the second UAV to perform cleaning tasks separately based on the time status of each contaminated task point, rather than simply initiating a uniform water rinse at a fixed time.

[0086] During the second UAV's water rinsing process, the ground station simultaneously acquires the second UAV's current location and the contaminated task points that have not yet been rinsed, and arranges the rinsing sequence according to the allowable rinsing time window for each contaminated task point. For contaminated task points that have entered the allowable rinsing time window and are close to the upper limit of the time window, the second UAV is given priority to rinse them, thereby reducing the possibility of the cleaning agent remaining on the contaminant surface for too long.

[0087] For example, the first drone sequentially sprays cleaning agent at the first, second, and third contaminated sites, with each site completing spraying at a different time. The ground station starts timing from each site's spray completion time. When the first contaminated site has entered its permitted rinsing window, while the second and third sites have not, the second drone prioritizes rinsing the first site with clean water. Subsequently, it continues rinsing based on the order in which the second and third sites entered their permitted rinsing windows.

[0088] When the second UAV arrives at a contaminated task point ahead of time, but the contaminated task point has not yet reached the allowable flushing time, the second UAV can temporarily suspend water spraying, or under the dispatch of the ground station, go to other contaminated task points that have entered the allowable flushing time window to perform flushing tasks, and return to perform water flushing after the contaminated task point meets the action time requirements.

[0089] If the second UAV fails to complete water rinsing at a contaminated area beyond the permitted rinsing time window due to insufficient water supply, flight path adjustments, or other reasons, the ground station will not directly control the second UAV to rinse the contaminated area. Instead, it will re-designate the contaminated area as the spraying point. The first UAV will spray cleaning agent onto the contaminated area again and record the spraying completion time, thereby regenerating the corresponding permitted rinsing time window. The second UAV will then complete the water rinsing within the new permitted rinsing time window.

[0090] After the second drone arrives at the designated pollution point, it can select the appropriate water spray nozzle based on the type of pollution. For stubborn pollutants such as bird droppings, a 0-degree direct spray nozzle or a 15-degree narrow-width spray nozzle can be used; for area-wide dust accumulation, a 25-degree medium-width spray nozzle can be used; and for large areas of floating dust, a 40-degree wide-width spray nozzle can be used. This allows the type of pollution to be simultaneously considered in controlling the cleaning agent's action time and selecting the subsequent water rinsing method.

[0091] After rinsing with clean water, the first UAV will re-inspect the corresponding area in accordance with the method described in the above embodiment. When the re-inspection result meets the preset cleaning requirements, the corresponding contaminated task point will be marked as task completed; if there are still residual contaminated areas that have not met the preset cleaning requirements, the area will be re-established as a contaminated task point, and the cleaning agent spraying, action time control, clean water rinsing and re-inspection process will be executed again.

[0092] This embodiment records the completion time of cleaning agent spraying for each contaminated task point, and the ground station establishes corresponding allowable rinsing time windows based on the spraying completion times of different contaminated task points. This ensures that the flight and rinsing tasks of the second UAV are directly constrained by the actual spraying status of the first UAV. Therefore, even if the first UAV continuously processes multiple contaminated areas, or if the second UAV returns for resupply or adjusts its flight path, the rinsing sequence can be rearranged according to the actual cleaning agent action time of each contaminated area. This reduces situations where the cleaning agent action time is insufficient or exceeds the appropriate action time, allowing the two UAVs to move beyond simple task division to form a collaborative cleaning process based on the status of the contaminated task points.

[0093] In summary, this invention first utilizes a first UAV to conduct visible light and infrared inspections of the photovoltaic modules. Based on the inspection data, the location, type, and area of ​​contaminants are determined. Then, the ground station selects either a precision cleaning mode or a non-discriminatory cleaning mode based on the proportion of contaminated area. In precision cleaning mode, the first UAV pre-treats stubborn local contaminants with foam cleaning agent. After an action time of 30 seconds to 3 minutes, a second UAV uses 0°, 15°, 25°, or 40° nozzles to rinse with clean water, depending on the type of contaminant. In non-discriminatory cleaning mode, both UAVs jointly perform large-area clean water spraying. After cleaning, the first UAV conducts a second inspection, and the results of this second inspection serve as the basis for determining whether supplementary cleaning is needed. This establishes a closed-loop operation process that seamlessly integrates inspection, contaminant analysis, mode selection, collaborative cleaning, replenishment, and cleaning quality re-inspection.

[0094] While the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the invention. Those skilled in the art can make various modifications and refinements without departing from the spirit and scope of the invention. Therefore, the scope of protection of the present invention shall be determined by the claims.

Claims

1. A photovoltaic precision cleaning method based on dual-UAV collaboration, characterized in that, Includes the following steps: S1: The first UAV performs an inspection of the target area of ​​the photovoltaic module and collects image data of the target area; S2: The ground station receives the image data and performs pollution identification and classification based on the image data to obtain pollution classification results; S3: The ground station determines the cleaning mode based on the pollution classification results and generates a corresponding cleaning task based on the determined cleaning mode. S4: When the cleaning mode is the precision cleaning mode, the first drone flies to the identified contaminated area and sprays cleaning agent onto the contaminated area; S5: After a preset action time, the second drone flies to the contaminated area that has been sprayed with the cleaning agent and performs a water rinse on the contaminated area; S6: After rinsing with clean water, the first UAV will perform a re-inspection of the cleaned contaminated area and collect re-inspection image data; S7: The ground station determines whether the cleaned contaminated area meets the preset cleaning requirements based on the re-inspection image data. When it is determined that the preset cleaning requirements have not been met, the ground station controls the first UAV and the second UAV to re-clean and re-inspect the contaminated area that has not met the preset cleaning requirements.

2. The photovoltaic precision cleaning method based on dual UAV collaboration according to claim 1, characterized in that, In step S1, the first drone collects image data of the photovoltaic module using a visible light camera and an infrared thermal imaging camera; in step S2, the location, type, and area of ​​pollutants are determined based on the image data, and the pollutants are classified into Class A pollution, Class B pollution, and Class C pollution, wherein Class A pollution is stubborn stains, Class B pollution is moderate pollution, and Class C pollution is light pollution; Class A pollution includes at least one of bird droppings, oil stains, and cement spots, Class B pollution includes regional dust accumulation, and Class C pollution includes large areas of floating dust.

3. The photovoltaic precision cleaning method based on dual UAV collaboration according to claim 2, characterized in that, In step S3, the ground station calculates the area ratio of the Class A contaminated area to the total contaminated area and the area ratio of the Class C contaminated area to the total contaminated area. When the area ratio of the Class A contaminated area to the total contaminated area is greater than 10%, the precision cleaning mode is activated. When the area ratio of the Class C contaminated area to the total contaminated area is greater than 80%, the indiscriminate cleaning mode is activated. In other cases, the operator selects the cleaning mode.

4. The photovoltaic precision cleaning method based on dual UAV collaboration according to claim 3, characterized in that, In the precision cleaning mode, the ground station plans the coordinated flight path of the first UAV and the second UAV according to the distribution of the contaminated area, so that the flight path of the first UAV passes through the Class A contaminated area first, and the time when the second UAV arrives at the corresponding contaminated area is timed relative to the time when the first UAV sprays the cleaning agent, with a preset action time interval; the preset action time is 30 seconds to 3 minutes.

5. The photovoltaic precision cleaning method based on dual UAV collaboration according to claim 4, characterized in that, The first drone uses a foam nozzle to spray the cleaning agent. The foam nozzle sprays the cleaning agent in foam form through an atomizing core assembly, and the spray range and coverage width of the foam are limited by a conical diffuser set at the nozzle. The second drone selects a clean water nozzle with different spray angles according to the type of pollution. For stubborn stains, a 0-degree direct spray nozzle or a 15-degree narrow spray nozzle is used. For large areas of light pollution, a 40-degree wide spray nozzle is used. For areas of dust accumulation, a 25-degree medium spray nozzle or a 40-degree wide spray nozzle is used.

6. The photovoltaic precision cleaning method based on dual UAV collaboration according to claim 5, characterized in that, During the operation of the first drone and / or the second drone, the battery level and liquid level of the corresponding drone are monitored in real time. When the battery level or liquid level is lower than a preset threshold, the corresponding drone is controlled to return to base to replenish the battery and / or liquid. After replenishment is completed, the unfinished cleaning task is continued. In step S7, whether the area of ​​residual stains in the cleaning area is less than the preset residual stain area is used as the basis for judging whether the cleaning meets the standard.

7. The photovoltaic precision cleaning method based on dual UAV collaboration according to claim 6, characterized in that, When the indiscriminate cleaning mode is activated, both the first and second drones are equipped with clean water nozzles. The ground station plans a large-area coverage route for the two drones based on the distribution of the photovoltaic modules to be cleaned. The first and second drones work in staggered areas along the large-area coverage route and jointly perform large-area clean water spray cleaning of the photovoltaic modules.

8. A photovoltaic precision cleaning system based on dual-UAV collaboration, used to implement the photovoltaic precision cleaning method based on dual-UAV collaboration as described in any one of claims 1 to 7, characterized in that, Includes the first drone, the second drone, and the ground station; The first UAV is equipped with an inspection payload and a replaceable cleaning payload, used to collect inspection image data of photovoltaic modules, spray cleaning agent on the polluted area according to the cleaning task issued by the ground station, and collect re-inspection image data after cleaning is completed. The second UAV is equipped with a water spraying assembly, which is used to rinse the contaminated area that has been sprayed with cleaning agent with water according to the cleaning task issued by the ground station; The ground station is communicatively connected to the first UAV and the second UAV respectively, and is equipped with a cleaning task scheduling module. The cleaning task scheduling module is used to identify and classify pollution based on the inspection image data, determine the cleaning mode based on the pollution classification result and schedule the first UAV and the second UAV to perform the corresponding cleaning task, and determine whether the cleaning meets the standard based on the re-inspection image data, and trigger a supplementary cleaning task when it is determined that the standard is not met.

9. The photovoltaic precision cleaning system based on dual UAV collaboration according to claim 8, characterized in that, The inspection payload of the first UAV includes a visible light camera and an infrared thermal imaging camera. The visible light camera has 2 megapixels and the infrared thermal imaging camera has a resolution of 640×512. The replaceable cleaning payload of the first UAV includes a cleaning agent foam nozzle. The cleaning agent foam nozzle includes an atomizing core assembly and a conical diffuser set at the nozzle, and is connected to the first UAV through a quick-release interface. The second UAV's water spray assembly includes a water tank, a high-pressure pump, a water nozzle assembly, and a two-degree-of-freedom gimbal. The water tank has a capacity of 30 to 50 liters and is equipped with an anti-sway structure. The high-pressure pump has an adjustable water pressure range of 0 to 12 MPa and a rated flow rate of not less than 10 L / min. The water nozzle assembly includes a 0-degree direct spray nozzle, a 15-degree narrow spray nozzle, a 25-degree medium spray nozzle, and a 40-degree wide spray nozzle. Each of the water nozzles is installed via a quick-release interface. The two-degree-of-freedom gimbal has a pitch range of -60 degrees to +35 degrees and a horizontal range of ±45 degrees.

10. The photovoltaic precision cleaning system based on dual UAV collaboration according to claim 8 or 9, characterized in that, The cleaning task scheduling module is configured as follows: Receive inspection image data collected by the first UAV, and obtain the location, type and area of ​​pollutants based on the inspection image data; Pollution is classified into three categories: Category A, Category B, and Category C. Category A pollution represents stubborn stains, Category B pollution represents moderate pollution, and Category C pollution represents light pollution. Calculate the area percentage of the Class A and Class C contaminated areas in all contaminated areas, and determine whether to activate the precision cleaning mode or the indiscriminate cleaning mode based on the area percentage. In the precision cleaning mode, a dual-drone collaborative route is planned, so that the first drone arrives at the contaminated area first to spray cleaning agent, and the second drone arrives at the corresponding contaminated area after a preset action time to perform water rinsing. In the indiscriminate cleaning mode, a dual-drone collaborative large-area coverage route is planned so that the first drone and the second drone work in staggered areas to jointly perform clean water spray cleaning. It also receives re-inspection image data collected by the first drone, determines whether the cleaning area meets the preset cleaning requirements based on the re-inspection image data, and generates a supplementary cleaning task for the unqualified contaminated area if the preset cleaning requirements are not met.

Images