Dirt distribution-based dynamic planning method for photovoltaic cleaning path of unmanned aerial vehicle

Abstract

The invention discloses an unmanned aerial vehicle photovoltaic cleaning path dynamic planning method based on smudginess distribution, relates to the technical field of unmanned aerial vehicle path dynamic planning, and aims to solve the technical problems of poor cleaning effect and low operation efficiency caused by single-dimensional planning in the existing method. Comprising the following steps: S1, a distributed data acquisition step: photovoltaic panel surface smudginess characteristic data, unmanned aerial vehicle cleaning equipment function adaptability data, photovoltaic panel array topographic data and unmanned aerial vehicle energy consumption characteristic data are synchronously acquired, and the smudginess characteristic data comprise smudginess adhesion degree, chemical properties and regional distribution density; the equipment function adaptability data contains the processing capacity and cross-regional operation adaptability of the cleaning part to different smudginess, and the energy consumption characteristic data contains energy consumption differences in different smudginess cleaning scenes and different terrains. The photovoltaic cleaning device has the advantages that the photovoltaic cleaning accuracy, efficiency and safety are improved, and continuous operation is guaranteed.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of unmanned aerial vehicle path dynamic programming, in particular to an unmanned aerial vehicle photovoltaic cleaning path dynamic programming method based on dirt distribution. BACKGROUND

[0002] Under the background of rapid development of new energy industry, the scale of photovoltaic power station as the core facility of clean energy continues to expand, which includes photovoltaic power stations in complex terrain (such as mountainous area, slope area). Because it can efficiently use idle land resources, it becomes an important construction type. However, this kind of photovoltaic power station faces significant challenges in the photovoltaic panel cleaning link: the dirt on the surface of the photovoltaic panel shows obvious regional differentiation characteristics (such as large differences in dirt adhesion, chemical properties and distribution density in different areas), and is affected by terrain conditions and regional environment (wind speed, humidity), the functional adaptability of unmanned aerial vehicle cleaning equipment, energy consumption state and the matching difficulty of operation path are greatly improved.

[0003] The current mainstream unmanned aerial vehicle photovoltaic cleaning path planning method does not construct a deep correlation system of "dirt characteristics-equipment function-regional environment-energy consumption state", but only carries out task allocation and path planning according to a single dimension (such as dirt density or fixed path). This leads to the problem that in actual operation, the function of the unmanned aerial vehicle does not match the characteristics of the dirt (such as being unable to effectively handle high adhesion dirt), the energy consumption of the equipment does not match the terrain conditions (such as abnormal energy consumption in steep slope area), and there is a lack of targeted obstacle avoidance and dynamic adjustment mechanism when multiple machines cooperate, etc. Ultimately, it results in poor cleaning effect, low operation efficiency and fast equipment wear, which is difficult to meet the needs of complex terrain photovoltaic power station for cleaning operation precision, stability and economy. In view of this, we propose an unmanned aerial vehicle photovoltaic cleaning path dynamic programming method based on dirt distribution. SUMMARY

[0004] The purpose of the present application is to provide an unmanned aerial vehicle photovoltaic cleaning path dynamic programming method based on dirt distribution, to solve the technical problems of poor cleaning effect and low operation efficiency caused by single dimension planning in the prior art.

[0005] To solve the above technical problems, the present application provides the following technical scheme: an unmanned aerial vehicle photovoltaic cleaning path dynamic programming method based on dirt distribution, comprising the following steps:

[0006] S1. Distributed data acquisition step: synchronously acquire the data of the dirt characteristics on the surface of the photovoltaic panel, the data of the functional adaptability of the unmanned aerial vehicle cleaning device, the data of the terrain of the photovoltaic panel array, and the data of the energy consumption characteristics of the unmanned aerial vehicle, wherein the dirt characteristics data include the adhesion degree, chemical properties, and regional distribution density of the dirt, the functional adaptability data include the processing capacity of the cleaning component for different dirt and the adaptability of the unmanned aerial vehicle to cross-region operation, and the energy consumption characteristics data include the energy consumption difference under different cleaning scenarios and different terrains, and all the data are classified according to the correlation dimension of "dirt-device-energy consumption-region";

[0007] S2. Intelligent task allocation step: based on the classified correlation data in S1, a four-dimensional dynamic decision matrix of "dirt characteristic adaptability-device function matching degree-energy consumption cost performance-region adaptability" is constructed, and an improved ant colony algorithm fusing "device capacity pre-evaluation + region adaptability pre-judgment" is adopted to cooperatively allocate the cleaning tasks and paths of each unmanned aerial vehicle;

[0008] S3. Dynamic cooperative control step: multi-unmanned aerial vehicle collision-free cooperation is realized, the dirt characteristic data in S1 and the task path and device adaptation strategy in S2 are corrected in real time through a three-feedback mechanism of "cleaning effect-device state-region environment", and a "resource pre-scheduling + task splitting + region cooperation" mode is adopted to process sudden dirt events.

[0009] Preferably, the S1 distributed data acquisition step specifically includes the following sub-steps:

[0010] S11. Dirt characteristic acquisition sub-step: the data of the dirt type, coverage range, adhesion degree, chemical properties, and regional distribution density are synchronously acquired through the multispectral sensor, pressure sensing probe, and image recognition unit integrated in each unmanned aerial vehicle, and a local dirt thermal map with "characteristics-region" double labels is generated;

[0011] S12. Device functional adaptability acquisition sub-step: the processing capacity of the cleaning component for different dirt characteristics and the adaptability of the unmanned aerial vehicle power system to different terrains are monitored in real time through the sensing unit integrated in each unmanned aerial vehicle, and a device "function-region" adaptability list is generated;

[0012] S13. Energy consumption characteristic acquisition sub-step: based on the pre-stored terrain information of the photovoltaic panel array, the corresponding relationship data of "cleaning action-terrain-energy consumption" of the unmanned aerial vehicle are combined in real time, and an energy consumption characteristic thermal map under different cleaning scenarios and different terrains is generated;

[0013] S14. Regional environment acquisition sub-step: the wind speed and humidity data of each region are acquired in real time through the fixed sensors deployed in the photovoltaic power station, and are used to correct the energy consumption data in S13 and the subsequent path stability parameters.

[0014] Preferably, the improvement of the traditional ant colony algorithm in the S2 intelligent task allocation step specifically includes the following sub-steps:

[0015] S21. Double adaptation pre-evaluation sub-step: based on the equipment "function-area" adaptability list generated in S12, the unmanned aerial vehicles with matching degree ≥ preset threshold for both target area terrain and dirt characteristics are screened out, and the unmanned aerial vehicles with insufficient adaptability are excluded to enter the path planning link;

[0016] S22. Heuristic factor weight adjustment sub-step: set "dirt characteristic adaptability" and "area adaptability" as core heuristic factors, and set "equipment function matching degree" and "energy consumption cost performance" as adjustment factors, and dynamically adjust the weight of each factor:

[0017] Increase the weight of "equipment cleaning strength matching degree" for high adhesion degree dirt area;

[0018] Increase the weight of "unmanned aerial vehicle terrain adaptability" for steep slope area;

[0019] Increase the weight of "equipment waterproof performance matching degree" for high humidity area;

[0020] The dynamic weight of each heuristic factor is calculated by the following algorithm formula:

[0021]

[0022] Wherein, W i represents the dynamic weight of the i-th heuristic factor, which is used to quantify the importance of the heuristic factor in task allocation decision-making, and the greater the weight value, the stronger the guiding effect on task allocation direction;

[0023] α represents the weight coefficient of adaptability, which is used to adjust the contribution proportion of the corresponding adaptability of the heuristic factor to the dynamic weight, and can be flexibly adjusted according to the priority requirements of different cleaning scenes;

[0024] F i represents the adaptability corresponding to the i-th heuristic factor, which includes dirt characteristic adaptability, area adaptability and equipment function adaptability, and is a quantitative index for measuring the matching degree of unmanned aerial vehicle and task demand;

[0025] β represents the weight coefficient of energy consumption characteristics, which is used to adjust the contribution proportion of energy consumption characteristics to the dynamic weight, and balances the adaptability and energy economy together with α;

[0026] E base is the energy consumption characteristic basic value of the unmanned aerial vehicle in the target area, and its reciprocal reflects the energy consumption economy;

[0027] S23. Double-mechanism pre-judgment sub-step: Based on historical cleaning data, current equipment status and regional environment data, the probability, effect and equipment loss degree of each unmanned aerial vehicle completing the task are pre-judged, and the task is preferentially assigned to the unmanned aerial vehicle with "high matching degree + high expected effect + low loss".

[0028] Preferably, the collision-free cooperation in the S3 dynamic cooperation control step specifically includes the following sub-steps:

[0029] S31. Complementary avoidance sub-step: Based on the real-time position data of each unmanned aerial vehicle, the equipment function status and the regional environment data, the "speed field obstacle avoidance + function-regional complementary avoidance" logic is adopted, when the paths of two unmanned aerial vehicles intersect, the path of the "high adhesion degree dirt special machine type" or "steep slope area special machine type" is preferentially guaranteed, and the other party adjusts the path and flight speed in combination with the wind speed data;

[0030] The path adjustment amount of the unmanned aerial vehicle is specifically calculated by the following algorithm formula:

[0031] ΔP=k×W func-area ·(V max -V curr );

[0032] Wherein, ΔP represents the path adjustment amount of the unmanned aerial vehicle, which can be embodied as a lateral offset distance and a height adjustment value, and is used to quantify the path parameters that the unmanned aerial vehicle needs to adjust for avoiding other unmanned aerial vehicles;

[0033] k represents a path adjustment coefficient, which is used to control the overall magnitude of the path adjustment amount, so as to avoid that the adjustment is too large to cause task delay or too small to achieve effective obstacle avoidance;

[0034] W func-area is the dynamic weight corresponding to the "function-region" adaptation degree, and the higher the weight value is, the stronger the adaptation of the unmanned aerial vehicle to the current regional task is, and other unmanned aerial vehicles need to make a larger path adjustment for it;

[0035] V max represents the maximum safe flight speed of the unmanned aerial vehicle, which is the upper limit of the safe flight speed that the unmanned aerial vehicle can reach under the current environment, and is determined by the unmanned aerial vehicle performance and the regional environment;

[0036] V curr represents the current flight speed of the unmanned aerial vehicle, i.e. the actual flight speed of the unmanned aerial vehicle when the paths intersect, and the difference between the maximum safe flight speed reflects the speed adjustment potential of the unmanned aerial vehicle;

[0037] S32. Fault emergency transfer sub-step: when a certain unmanned aerial vehicle triggers an equipment fault warning, immediately search for a surrounding "function-region dual adaptation" replacement unmanned aerial vehicle, split and transfer the unfinished tasks of the faulty unmanned aerial vehicle according to "dirt characteristics-region", and ensure task continuity and matching degree unchanged;

[0038] S33. Wind speed adaptive adjustment sub-step: dynamically adjusting the flight height and path spacing of the UAV in the area where the wind speed is greater than or equal to the preset threshold, and increasing the obstacle avoidance redundancy.

[0039] Preferably, the three feedback corrections in the S3 dynamic cooperative control step specifically include the following sub-steps:

[0040] S34. Multi-data receiving sub-step: receiving the "post-washing image data + equipment state change data + real-time regional environment data" completed by each UAV after washing;

[0041] S35. Cause determination and correction sub-step: determining the problem cause by analyzing the dirt residue amount through image data, combining equipment state data and regional environment data:

[0042] If the dirt residue amount exceeds the standard, the equipment is normal, and the environment is normal, the path density of the area is corrected;

[0043] If the dirt residue amount exceeds the standard, the equipment is abnormal, and the environment is normal, the "function-adapted" UAV is re-assigned;

[0044] If the dirt residue amount meets the standard, the equipment is abnormal, and the humidity exceeds the standard, the equipment "function-area" adaptability list is updated;

[0045] The path density correction coefficient is calculated by the following algorithm formula:

[0046]

[0047] Wherein, K d represents the path density correction coefficient, which is used to quantify the proportion of the original path density that needs to be adjusted. A coefficient greater than 1 indicates that the path density needs to be increased, and a coefficient close to 1 indicates that the original path density is basically reasonable;

[0048] γ represents the correction coefficient adjustment factor, which is used to control the influence degree of the path adjustment amount on the correction coefficient, so as to avoid excessive correction caused by single adjustment amount fluctuation;

[0049] ΔP is the UAV path adjustment amount, and its absolute value reflects the deviation degree of the original path planning and the actual scene (such as obstacle avoidance demand and cleaning demand);

[0050] P std represents the standard path spacing of the target area, i.e. the reasonable path spacing preset based on the dirt characteristics and terrain conditions of the area, which is the basis for judging whether the path density is reasonable;

[0051] S36. Data updating sub-step: based on the determination result of S35, the "characteristics-area" double label of the global dirt thermodynamic map, the energy consumption characteristic thermodynamic map, and the equipment "function-area" adaptability list are updated in real time.

[0052] Preferably, the burst contamination event processing in the S3 dynamic cooperative control step specifically includes the following sub-steps:

[0053] S37. Emergency level triggering sub-step: Real-time monitoring of the surface contamination concentration of the photovoltaic panel, the characteristic change and the regional environment data, when the local area appears sudden increase in contamination concentration, special characteristics or accompanied by extreme weather, triggering the corresponding level of emergency response, marking the "emergency contamination characteristics-emergency regional environment" label;

[0054] S38. Priority scheduling sub-step: From the currently working drones, select the drones with "function adaptation emergency contamination characteristics+region adaptation emergency environment+small remaining task amount", after transferring the unfinished ordinary tasks of the drones to the surrounding adaptive drones according to "region", schedule the drones to perform emergency tasks;

[0055] S39. Sub-stage processing sub-step: If there is no immediately schedulable adaptive drone, start "off-site adaptive drone pre-scheduling", and at the same time, split the emergency area task into "temporary protection cleaning+deep precise cleaning", first by the nearby ordinary drones to perform temporary protection cleaning, and then by the adaptive drones to perform deep precise cleaning after arriving on the scene.

[0056] Preferably, the S3 dynamic cooperative control step further includes:

[0057] S310. Loss prediction sub-step: Based on real-time device state data, historical loss data and current task characteristics, establish a device loss prediction model to calculate the estimated loss value of each drone after completing the current task;

[0058] S311. Task transfer warning sub-step: When the estimated loss value of a drone is greater than or equal to the preset safety threshold, trigger "task transfer warning", search for surrounding adaptive drones and gradually transfer the unfinished tasks, while generating a device maintenance reminder;

[0059] S312. Loss balancing sub-step: For drones that have been handling high-loss tasks for a long time, adjust their subsequent task allocation strategy to increase the proportion of low-loss tasks, achieving device loss balancing.

[0060] Preferably, the S3 dynamic cooperative control step further includes:

[0061] S313. Extreme environment triggering sub-step: When S14 detects an extreme environment, immediately trigger path emergency adjustment;

[0062] S314. Path optimization sub-step: Based on extreme environment data, optimize the flight path of all drones in the region, including shortening the flight distance, reducing the flight height, avoiding the windward area, and temporarily suspending high-risk cleaning actions;

[0063] S315. Safety scheduling sub-step: If the extreme environment duration is greater than or equal to the preset time length, the unmanned aerial vehicle in the region is scheduled to a preset safety region, the task is suspended, and a restart scheme after the environment is recovered is generated. After the environment is recovered, the task path is restarted or adjusted based on the dirty residual data and the equipment state.

[0064] Preferably, the method further comprises:

[0065] S4. Visualization monitoring and intervention step: a digital twin interface associated with "dirt-equipment-path-region" is constructed, the matching state of each unmanned aerial vehicle and the dirt scene, the region environment, the task iteration process and the energy consumption optimization effect are displayed in real time, and precise intervention and cross-region resource scheduling of the operation and maintenance personnel are supported based on the associated data;

[0066] The S4 visualization monitoring and intervention step further comprises:

[0067] S41. Resource state identification sub-step: based on the associated data of "dirt density-equipment quantity-equipment adaptation degree" of each region displayed by the digital twin interface, the resource idle area and the resource shortage area are identified;

[0068] S42. Scheduling instruction execution sub-step: the operation and maintenance personnel initiate cross-region resource scheduling instructions through the visualization interface, and the system plans the cross-region scheduling path of the idle equipment based on the "equipment-region dual adaptation" principle, and adjusts the task allocation scheme of the target region;

[0069] S43. Scheduling path optimization sub-step: during the cross-region scheduling process, the scheduling path is dynamically optimized in combination with the region environment data along the way to ensure safe and efficient arrival of the equipment.

[0070] Preferably, the S4 visualization monitoring and intervention step further comprises:

[0071] S44. Data storage sub-step: store the "initial dirty data-equipment information-cleaning path-post-cleaning data-equipment loss data" of each cleaning in each region to form a cleaning effect traceability library;

[0072] S45. Correlation analysis sub-step: based on the traceability library data, regularly generate a "dirt characteristics-equipment adaptation-cleaning effect" correlation analysis report to identify the optimal adaptation combination;

[0073] S46. Strategy optimization sub-step: the optimal adaptation combination is fed back to the S2 intelligent task allocation step as a reference basis for subsequent task allocation, continuously optimizing the path planning and equipment matching accuracy.

[0074] Compared with the prior art, the present application has the following advantages:

[0075] 1. The application realizes the coordinated and accurate allocation of unmanned aerial vehicle cleaning tasks and paths by synchronously collecting multi-dimensional data of "dirt characteristics-equipment functions-regional environment-energy consumption state" and establishing a correlation classification system, and constructing a four-dimensional dynamic decision matrix combined with an improved ant colony algorithm, ensuring that the functions and energy consumption state of the unmanned aerial vehicle are highly matched with the dirt characteristics and terrain environment of the target region, and effectively solving the core problems of poor cleaning effect and low operation efficiency caused by single-dimensional planning in existing methods.

[0076] 2. The application also introduces a "cleaning effect-equipment state-regional environment" three feedback mechanism to correct the dirt data, task path and equipment adaptation strategy in real time, and processes sudden dirt events through the "resource pre-scheduling + task splitting + regional cooperation" mode, avoiding operation interruption or adaptation deviation caused by dynamic scene changes (such as abnormal dirt residue and sudden dirt), further solving the problem of static planning unable to cope with dynamic scenes in the core beneficial effect, and ensuring the continuity and stability of cleaning operations.

[0077] 3. The application also realizes dynamic control of equipment wear and tear and safe operation in extreme scenarios through an equipment wear prediction model and an extreme environment path optimization strategy, predicts wear and tear based on equipment state and task characteristics, transfers tasks in advance and reminds maintenance; real-time adjustment of path and operation mode for extreme environment, scheduling equipment to safe area, further solving the problem of not considering long-term wear and tear of equipment and safety in extreme environment in the core and further beneficial effect, prolonging the service life of equipment, and improving the economy and safety of the overall operation. BRIEF DESCRIPTION OF DRAWINGS

[0078] Figure 1 The figure is a schematic diagram of the method of the application. DETAILED DESCRIPTION

[0079] To facilitate those skilled in the art to understand the technical solutions of the application, the technical solutions of the application will be further described in conjunction with the drawings of the specification.

[0080] In Example 1, as shown, the application provides a dynamic planning method for unmanned aerial vehicle photovoltaic cleaning path based on dirt distribution, including the following steps: Figure 1 S1. Distributed data acquisition step: synchronously collect photovoltaic panel surface dirt characteristic data, unmanned aerial vehicle cleaning equipment function adaptability data, photovoltaic panel array terrain data and unmanned aerial vehicle energy consumption characteristic data, the dirt characteristic data includes dirt adhesion, chemical properties and regional distribution density, the equipment function adaptability data includes the processing capacity of cleaning components for different dirt and cross-regional operation adaptability, and the energy consumption characteristic data includes the energy consumption difference under different dirt cleaning scenarios and different terrains, and all data are classified according to the "dirt-equipment-energy consumption-region" correlation dimension;

[0081] ​

[0082] S2. Intelligent task allocation step: based on the classified associated data in S1, a four-dimensional dynamic decision matrix of "dirt characteristic adaptation degree-equipment function matching degree-energy cost performance-region adaptation degree" is constructed, an improved ant colony algorithm combining "equipment capacity pre-evaluation + regional adaptability pre-judgment" is used to cooperatively allocate the cleaning tasks and paths of each unmanned aerial vehicle, and the unmanned aerial vehicle equipment function, regional adaptability and target region dirt characteristics, terrain conditions are accurately matched;

[0083] S3. Dynamic cooperative control step: multi-unmanned aerial vehicle collision-free cooperation is realized, the dirt characteristic data in S1 and the task path and equipment adaptation strategy in S2 are corrected in real time through a three-feedback mechanism of "cleaning effect-equipment state-region environment", and a "resource pre-scheduling + task splitting + regional cooperation" mode is used to handle sudden dirt events, so as to avoid global task interruption and regional resource idling.

[0084] In the embodiment of the present application, the S1 distributed data acquisition step specifically includes the following sub-steps:

[0085] S11. Dirt characteristic acquisition sub-step: through the multi-spectral sensor, pressure sensing probe and image recognition unit integrated in each unmanned aerial vehicle, the dirt type, coverage range, adhesion degree, chemical property and regional distribution density data are synchronously acquired, and a local dirt thermal map with "characteristic-region" double labels is generated;

[0086] S12. Equipment function adaptability acquisition sub-step: through the sensing unit integrated in each unmanned aerial vehicle, the processing capacity of cleaning components (brush head material, spray pressure mode, cleaning agent compatibility) for different dirt characteristics and the adaptability of unmanned aerial vehicle power system to different terrains are monitored in real time, and an equipment "function-region" adaptability list is generated;

[0087] S13. Energy consumption characteristic acquisition sub-step: based on the pre-stored photovoltaic panel array terrain information, the corresponding relationship data of "cleaning action-terrain-energy consumption" of the unmanned aerial vehicle are combined in real time, and an energy consumption characteristic thermal map under different cleaning scenes and different terrains is generated;

[0088] The energy consumption characteristic basic value of the unmanned aerial vehicle in the target region is calculated by the following algorithm formula, which provides data support for subsequent energy cost performance calculation:

[0089] E base =f(A t ,T r ,C w );

[0090] Wherein, E base represents the energy consumption characteristic basic value of the unmanned aerial vehicle in the target region, which is a basic quantitative index for measuring the energy consumption level of the unmanned aerial vehicle when performing cleaning tasks in the region;

[0091] f(·) represents the energy consumption characteristic mapping function, which is used to convert the three input parameters of cleaning action, terrain type and dirt coverage into energy consumption characteristic basic value through preset correlation rules (such as weighted calculation, classification mapping);

[0092] A t represents the cleaning action type of the unmanned aerial vehicle in the area, covering different cleaning methods such as spraying, brushing and high-pressure washing, and the energy consumption of different actions is different;

[0093] T r represents the terrain type of the target area, including flat, gentle slope and steep slope, and the terrain difference will affect the flight resistance and power output of the unmanned aerial vehicle, and then affect the energy consumption;

[0094] C w represents the dirt coverage of the target area, that is, the coverage ratio of dirt on the surface of the photovoltaic panel, and the higher the coverage, the longer the cleaning time or the stronger the cleaning intensity, and the energy consumption increases accordingly;

[0095] The formula is used to calculate the energy consumption characteristic basic value of the unmanned aerial vehicle in the target area, and the core logic is to establish the correlation between the three and the energy consumption basic value by combining the specific cleaning action of the unmanned aerial vehicle in the target area, the terrain type of the area and the dirt coverage, and to generate energy consumption characteristic basic data through function mapping, to provide quantitative basis for subsequent judgment of energy consumption performance and optimization of task allocation, and to avoid the deviation caused by relying on single factor to estimate energy consumption;

[0096] The energy consumption characteristic basic value calculated by the formula can accurately associate the key influencing factors in the cleaning scene, avoid the inaccuracy caused by ignoring the scene difference in traditional energy consumption estimation, provide reliable data support for the energy consumption performance evaluation in the subsequent task allocation, help to preferentially select the unmanned aerial vehicle with higher matching degree of energy consumption and task demand, reduce invalid energy consumption from the source, and improve the energy utilization efficiency of the whole cleaning operation;

[0097] S14. Area environment acquisition sub-step: real-time acquisition of wind speed and humidity data of each area through fixed sensors deployed in the photovoltaic power station, for correcting the energy consumption data in S13 and the subsequent path stability parameters.

[0098] In the embodiments of the present application, the improvement of the traditional ant colony algorithm in the S2 intelligent task allocation step specifically includes the following sub-steps:

[0099] S21. Double adaptation pre-evaluation sub-step: based on the "function-area" adaptability list generated in S12, the unmanned aerial vehicles with matching degree of target area terrain and dirt characteristics both greater than or equal to a preset threshold are selected, and the unmanned aerial vehicles with insufficient adaptability are excluded to enter the path planning link;

[0100] S22. Heuristic factor weight adjustment sub-step: Set "dirt property fitness" and "region fitness" as core heuristic factors, and set "device function matching degree" and "energy consumption cost performance" as adjustment factors, and dynamically adjust the weight of each factor:

[0101] Increase the weight of "device cleaning strength matching degree" for high adhesion degree dirt area;

[0102] Increase the weight of "unmanned aerial vehicle terrain adaptability" for steep slope area;

[0103] Increase the weight of "device waterproof performance matching degree" for high humidity area;

[0104] The dynamic weight of each heuristic factor is calculated by the following algorithm formula:

[0105]

[0106] Wherein, W i represents the dynamic weight of the i-th heuristic factor, which quantifies the importance of the heuristic factor in task allocation decision-making, and the greater the weight value, the stronger the guiding role in task allocation direction;

[0107] α represents the weight coefficient of the fitness, which is used to adjust the contribution proportion of the fitness corresponding to the heuristic factor to the dynamic weight, and can be flexibly adjusted according to the priority requirements of different cleaning scenes;

[0108] F i represents the fitness corresponding to the i-th heuristic factor, which includes dirt property fitness, region fitness, and device function fitness, and is a quantitative index for measuring the matching degree of unmanned aerial vehicle and task demand;

[0109] β represents the weight coefficient of energy consumption characteristics, which is used to adjust the contribution proportion of energy consumption characteristics to the dynamic weight, and balances the adaptability and energy economy together with α;

[0110] E base is the energy consumption characteristic basic value of the unmanned aerial vehicle in the target area, and its reciprocal reflects the energy consumption economy (the lower the energy consumption, the larger the reciprocal, and the stronger the positive contribution to the dynamic weight);

[0111] This formula is used to calculate the dynamic weight of each heuristic factor. The core logic is: taking the fitness corresponding to the heuristic factor (such as dirt property fitness, region fitness) as the core, combining the energy consumption characteristic basic value (the lower the energy consumption, the larger the reciprocal), adjusting the contribution degree of the two to the dynamic weight through the weight coefficient, realizing the coordinated consideration of adaptability and energy economy, and avoiding one-sided decision-making of only focusing on adaptability or only focusing on energy consumption;

[0112] The formula deeply couples the adaptability and energy consumption economy through dynamic weight calculation, solves the problem of fixed weight in traditional task allocation which cannot adapt to scene changes, can automatically adjust the importance of each heuristic factor according to the dirt characteristics, terrain conditions and energy consumption demand of different areas, guide the task allocation to tilt towards the direction of 'high adaptability + low energy consumption', ensure the cleaning effect (high adaptability) and control the energy consumption cost (good energy consumption economy), and improve the overall rationality and economy of task allocation.

[0113] S23. Double mechanism prediction substep: based on historical cleaning data, current equipment state and regional environment data, the probability, effect and equipment loss degree of each unmanned aerial vehicle completing the task are predicted, and the task is preferentially allocated to the unmanned aerial vehicle with 'high matching degree + high expected effect + low loss'.

[0114] In the embodiment of the application, the collision-free cooperation in the S3 dynamic cooperative control step specifically includes the following substeps:

[0115] S31. Complementary avoidance substep: based on real-time position data of each unmanned aerial vehicle, equipment function state and regional environment data, the'velocity field obstacle avoidance + function-region complementary avoidance' logic is adopted, when the paths of two unmanned aerial vehicles intersect, the path of the 'high adhesion degree dirt special machine type' or'steep slope region special machine type' is preferentially ensured, and the other party adjusts the path and flight speed in combination with the wind speed data;

[0116] The path adjustment amount of the unmanned aerial vehicle is specifically calculated by the following algorithm formula:

[0117] ΔP=k×W func-area ·(V max -V curr );

[0118] Wherein, ΔP represents the path adjustment amount of the unmanned aerial vehicle, which can be embodied as a lateral offset distance and a height adjustment value, and is used to quantify the path parameters that the unmanned aerial vehicle needs to adjust for avoiding other unmanned aerial vehicles;

[0119] K represents a path adjustment coefficient, which is used to control the overall magnitude of the path adjustment amount, so as to avoid that the adjustment is too large to cause task delay or too small to realize effective obstacle avoidance;

[0120] W func-area is a dynamic weight corresponding to the 'function-region' adaptability, and the higher the weight value is, the stronger the adaptability of the unmanned aerial vehicle to the current regional task is, and other unmanned aerial vehicles need to make larger path adjustment for it;

[0121] V max represents the maximum safe flight speed of the unmanned aerial vehicle, which is the upper limit of the safe flight speed that the unmanned aerial vehicle can reach under the current environment, and is determined by the unmanned aerial vehicle performance and the regional environment;

[0122] Vcurr represents the current flight speed of the UAV, that is, the actual flight speed of the UAV when the paths intersect, and the difference between the maximum safe flight speed reflects the speed adjustment potential of the UAV;

[0123] The formula is used to calculate the path adjustment amount of the UAV, and the core logic is: taking the dynamic weight of the "function-region" adaptation degree as the core basis (the higher the weight, the stronger the adaptation of the UAV to the region task, and the more it needs to prioritize the path), combining the difference between the current flight speed of the UAV and the maximum safe flight speed (the greater the difference, the more the adjustment space), and controlling the amplitude of the adjustment amount through an adjustment coefficient, realizing orderly avoidance when the paths intersect, avoiding collision and protecting the operation efficiency of the highly adapted UAV;

[0124] The path adjustment amount calculated by the formula can realize differentiated avoidance based on the adaptation difference of the UAV to the task, avoid the problem of reduced operation efficiency of the highly adapted UAV caused by traditional undifferentiated avoidance, ensure collision-free safety during multi-UAV cooperative operation, and preferentially maintain the operation continuity of the UAV with strong adaptability, reduce task interruption or path redundancy caused by avoidance, and improve the overall operation efficiency and safety of multi-UAV cooperative cleaning;

[0125] S32. Fault emergency transfer sub-step: when a UAV triggers a device fault warning, immediately retrieve a replacement UAV with "function-region dual adaptation" in the surrounding area, split and transfer the unfinished task of the faulty UAV according to "dirt characteristics-region", and ensure task continuity and unchanged matching degree;

[0126] S33. Wind speed adaptation adjustment sub-step: for the region where the wind speed is greater than or equal to the preset threshold, dynamically adjust the flight height and path distance of the UAV in the region, and increase the obstacle avoidance redundancy.

[0127] In the embodiments of the present application, the three feedback corrections in the S3 dynamic cooperative control step specifically include the following sub-steps:

[0128] S34. Multi-data receiving sub-step: receiving "post-cleaning image data + device state change data + real-time region environment data" after each UAV completes cleaning;

[0129] S35. Cause determination and correction sub-step: analyzing the dirt residue amount through image data, and determining the problem cause in combination with device state data and region environment data:

[0130] If the dirt residue amount exceeds the standard, the device is normal, and the environment is normal, the path density of the region is corrected;

[0131] If the dirt residue amount exceeds the standard, the device is abnormal, and the environment is normal, the "function adaptation" UAV is re-assigned;

[0132] If the residual amount of dirt meets the standard, the equipment is abnormal, and the humidity exceeds the standard, update the equipment "function-area" adaptability list;

[0133] The path density correction coefficient is calculated by the following algorithm formula:

[0134]

[0135] Wherein, K d represents the path density correction coefficient, which is used to quantify the proportion of the original path density that needs to be adjusted. A coefficient greater than 1 indicates that the path density needs to be increased (such as encrypting the path to improve cleaning coverage), and a coefficient close to 1 indicates that the original path density is basically reasonable.

[0136] γ represents the correction coefficient adjustment factor, which is used to control the influence degree of path adjustment amount on the correction coefficient, to avoid over-correction caused by single adjustment amount fluctuation;

[0137] ΔP is the path adjustment amount of the unmanned aerial vehicle, and its absolute value reflects the deviation degree of the original path planning and the actual scene (such as obstacle avoidance demand, cleaning demand);

[0138] P std represents the standard path spacing of the target area, that is, the reasonable path spacing preset based on the dirt characteristics and terrain conditions of the area, which is the basis for judging whether the path density is reasonable;

[0139] The formula is used to calculate the path density correction coefficient. The core logic is: based on the absolute value of the path adjustment amount of the unmanned aerial vehicle (the larger the adjustment amount, the greater the matching deviation of the original path planning and the actual scene), combined with the standard path spacing of the target area, the influence of the two on the correction coefficient is adjusted through the correction coefficient to generate a path density correction index, which is used to guide subsequent path optimization, so that the path density is more matched with the actual cleaning demand and obstacle avoidance demand;

[0140] Through the calculation of the path density correction coefficient, the formula can reversely optimize the path planning based on the historical path adjustment situation, solve the problem of fixed traditional path density that cannot be dynamically adjusted according to the actual scene, make the path density meet the cleaning coverage rate demand (avoid incomplete cleaning due to too sparse path), and adapt to the obstacle avoidance demand (avoid increasing the collision risk due to too dense path), realize the dynamic optimization of path planning, and improve the balance between cleaning effect and operation safety;

[0141] S36. Data updating sub-step: based on the determination result of S35, real-time update the "characteristic-area" double label of the global dirt thermal map, the energy consumption characteristic thermal map and the equipment "function-area" adaptability list.

[0142] In the embodiments of the present application, the sudden dirt event processing in the S3 dynamic cooperative control step specifically includes the following sub-steps:

[0143] S37. Emergency level triggering sub-step: Real-time monitoring of the concentration of dirt on the surface of the photovoltaic panel, the change of characteristics and regional environmental data, when the local area appears sudden increase of dirt concentration, special characteristics or accompanied by extreme weather, trigger the corresponding level of emergency response, mark the "emergency dirt characteristics-emergency regional environment" label;

[0144] The emergency response level is calculated by the following algorithm formula:

[0145]

[0146] Wherein, L emerg represents the emergency response level, which is a quantitative index to measure the emergency degree of the dirt outbreak event, the higher the level, the more urgent the coping strategy (such as priority scheduling, phased processing);

[0147] δ represents the level influence coefficient of dirt concentration, which is used to adjust the contribution proportion of the difference of dirt concentration to the emergency degree, and reflects that the dirt concentration is the core factor for judging the emergency level;

[0148] C curr represents the current regional dirt concentration, that is, the real-time dirt concentration level of the region after the sudden dirt event occurs;

[0149] C std represents the standard dirt concentration of the region, that is, the reasonable dirt concentration threshold of the region under normal circumstances, which is the basis for judging whether it belongs to sudden dirt;

[0150] ∈ represents the level influence coefficient of path density, which is used to adjust the contribution proportion of the path density correction coefficient to the emergency degree, and reflects the influence of the deviation of the original path planning on the difficulty of subsequent emergency treatment;

[0151] K d is the path density correction coefficient, which reflects the deviation degree of the original path planning and the actual scene;

[0152] The formula is used to calculate the emergency response level, and the core logic is: taking the ratio of the current regional dirt concentration to the standard dirt concentration (the larger the ratio, the higher the degree of dirt outbreak) as the core, combining the path density correction coefficient (the larger the correction coefficient, the greater the deviation of the original path planning, which needs higher emergency level to cope with), adjusting the contribution of the two to the emergency degree through the level influence coefficient, generating a quantitative emergency response level, providing a grading basis for subsequent emergency scheduling, avoiding over or insufficient emergency response;

[0153] The emergency response level calculated by the formula can comprehensively consider the degree of pollution burst and the deviation of the original path planning, realize accurate grading of emergency response, avoid resource waste (such as using high-level resources for low emergency events) or insufficient response (such as using low-level resources for high emergency events) caused by traditional "one-size-fits-all" emergency response, provide clear priority basis for subsequent emergency dispatch, ensure efficient matching of emergency resources and emergency demand, and improve the processing efficiency and effect of sudden pollution events.

[0154] S38. Priority dispatch sub-step: From the currently working unmanned aerial vehicle, select the unmanned aerial vehicle with the functions of "function adaptation emergency pollution characteristics + region adaptation emergency environment + small remaining task amount", and after the unmanned aerial vehicle is selected, the unfinished ordinary task of the unmanned aerial vehicle is transferred to the surrounding adaptive unmanned aerial vehicle according to the "region", and the unmanned aerial vehicle is dispatched to execute the emergency task;

[0155] S39. Sub-step of processing in stages: if there is no immediately dispatchable adaptive unmanned aerial vehicle, start "off-site adaptive unmanned aerial vehicle pre-dispatching", and at the same time, the emergency area task is divided into "temporary protection cleaning + deep precise cleaning", the temporary protection cleaning is first executed by the nearby ordinary unmanned aerial vehicle, and the deep precise cleaning is executed by the adaptive unmanned aerial vehicle after arriving at the scene.

[0156] In the embodiment of the application, the S3 dynamic cooperative control step further comprises:

[0157] S310. Loss prediction sub-step: based on real-time device state data, historical loss data and current task characteristics, a device loss prediction model is established, and an estimated loss value of each unmanned aerial vehicle after completing the current task is calculated;

[0158] The estimated loss value is calculated by the following algorithm formula:

[0159] W pre =μ×T task ·(1+ζ×L emerg );

[0160] Wherein, W pre represents the estimated loss value of the unmanned aerial vehicle after completing the current task, which is a quantitative index for measuring the loss degree of the device after the current task is completed, and is used to judge whether the device still has the ability to execute subsequent tasks;

[0161] μ represents a device basic loss coefficient, which is determined by the material and performance of the cleaning parts (such as brush head and motor) of the unmanned aerial vehicle, and is the basic loss rate of the device per unit time;

[0162] T task represents the current task expected duration, that is, the estimated time required for the unmanned aerial vehicle to complete the current cleaning task, the longer the duration, the longer the device continues to work, and the more the loss accumulates;

[0163] ζ represents the influence coefficient of emergency level on loss, used to adjust the contribution proportion of emergency response level to loss increment, and the emergency task usually accompanies higher intensity operation, and the loss increment is higher than that of the regular task;

[0164] L emerg is an emergency response level, reflecting the emergency intensity of the current task, and the higher the level, the greater the operation intensity and the faster the equipment loss;

[0165] The formula is used to calculate the estimated loss value of the unmanned aerial vehicle after completing the current task, and the core logic is: taking the product of the device basic loss coefficient and the predicted duration of the current task as the basis (the longer the duration, the greater the basic loss), combining the emergency response level (the higher the level, the greater the emergency task intensity, and the higher the loss increment), adjusting the loss increment proportion through the emergency level influence coefficient, generating a quantitative estimated loss value, providing a basis for determining whether the device needs to hand over the task in advance and performing maintenance, and avoiding resource waste caused by excessive loss or early maintenance of the device;

[0166] The estimated loss value calculated by the formula can accurately predict the loss state of the device after the current task, solve the problem that the traditional device maintenance relies on a fixed cycle and cannot be dynamically adjusted according to the task intensity, avoid the impact of device failure on operation continuity caused by not discovering the high loss state in time, avoid the resource idling caused by excessive maintenance (such as the device still has the ability but is offline in advance), realize dynamic control of device loss, prolong the service life of the device, and ensure the stable development of cleaning operation;

[0167] S311. Task handover warning sub-step: when the estimated loss value of a certain unmanned aerial vehicle is greater than or equal to a preset safety threshold, a "task handover warning" is triggered, surrounding suitable unmanned aerial vehicles are searched, and the unfinished task is gradually handed over, and a device maintenance reminder is generated;

[0168] S312. Loss balancing sub-step: for the unmanned aerial vehicle that has been handling high loss tasks for a long time, the subsequent task allocation strategy is adjusted, and the proportion of low loss tasks is increased, so as to realize the loss balancing of the device.

[0169] In the embodiment of the application, the method further comprises:

[0170] S4. Visual monitoring and intervention step: a digital twin interface associated with "dirt-device-path-region" is constructed, the matching state of each unmanned aerial vehicle and the dirt scene, the region environment, the task iteration process and the energy consumption optimization effect are displayed in real time, and precise intervention and cross-region resource scheduling based on associated data are supported for operation and maintenance personnel;

[0171] The S4 visual monitoring and intervention step further comprises:

[0172] S41. Resource state identification sub-step: based on the "dirt density-equipment quantity-equipment adaptation degree" correlation data of each area displayed on the digital twin interface, identify the resource idle area and the resource shortage area;

[0173] S42. Dispatching instruction execution sub-step: the operation and maintenance personnel initiates a cross-area resource dispatching instruction through the visual interface, and the system plans a cross-area dispatching path for the idle equipment based on the "equipment-area dual adaptation" principle, while adjusting the task allocation scheme of the target area;

[0174] The path priority of cross-area dispatching is calculated through the following algorithm formula:

[0175]

[0176] Wherein, P disp represents the path priority of cross-area dispatching, and the larger the priority value is, the more suitable the idle equipment is for being dispatched to the target area to perform tasks;

[0177] η represents the priority coefficient of distance, which is used to adjust the contribution proportion of distance factor to dispatching priority, and the closer the distance is, the shorter the dispatching time is and the lower the cost is;

[0178] D dist represents the straight-line distance between the idle equipment and the target area, which is an important factor for measuring dispatching cost and efficiency, and the farther the distance is, the greater the dispatching difficulty and time are usually;

[0179] θ represents the priority coefficient of equipment wear, which is used to adjust the contribution proportion of equipment wear factor to dispatching priority, and the lower the wear of the equipment is, the stronger the stability and continuity of subsequent task execution are;

[0180] W pre is the estimated wear value of the idle equipment, and its reciprocal reflects the availability of the equipment (the lower the wear is, the larger the reciprocal is, and the higher the availability of the equipment is);

[0181] The formula is used to calculate the path priority of cross-area dispatching, and the core logic is: taking the reciprocal of the straight-line distance between the idle equipment and the target area (the closer the distance is, the larger the reciprocal is, and the higher the priority basis is) as the basis, combining the reciprocal of the estimated wear value of the idle equipment (the lower the wear value is, the larger the reciprocal is, and the higher the availability of the equipment is), adjusting the contribution of the two to the dispatching priority through the priority coefficient, generating a quantitative path priority, providing a basis for selecting the optimal equipment for cross-area dispatching, and avoiding the selection of equipment with too far distance or too high wear, which leads to low efficiency or failure risk;

[0182] The path priority calculated by the formula can comprehensively consider the distance cost and device availability of cross-regional scheduling, realize optimal matching of scheduling resources, avoid the problem that the traditional scheduling only focuses on distance and ignores device state (such as selecting a near but high-loss device to cause a failure in the middle) or only focuses on device state and ignores distance (such as selecting a low-loss but too far device to cause scheduling delay), ensure that the scheduled device can quickly reach the target region and has the ability to stably perform the task, improve the efficiency and reliability of cross-regional resource scheduling, and balance the resource distribution of each region.

[0183] S43. Scheduling path optimization sub-step: during the cross-regional scheduling process, the scheduling path is dynamically optimized in combination with the regional environment data along the way to ensure safe and efficient arrival of the device.

[0184] In the embodiments of the present application, the S4 visualization monitoring and intervention step further comprises:

[0185] S44. Data storage sub-step: store the "dirt initial data-device information-cleaning path-post-cleaning data-device loss data" of each cleaning in each region to form a cleaning effect traceability library;

[0186] S45. Correlation analysis sub-step: based on the traceability library data, regularly generate a "dirt characteristics-device adaptation-cleaning effect" correlation analysis report to identify the optimal adaptation combination;

[0187] The effect score of the adaptation combination is specifically calculated by the following algorithm formula:

[0188] S adapt =λ×E clean +ξ×P disp ;

[0189] Wherein, S adapt represents the effect score of the adaptation combination, which is a quantitative index for measuring the overall effect of the "dirt characteristics-device-scheduling path" combination, and the higher the score, the better the overall performance of the combination;

[0190] λ represents the score coefficient of the cleaning effect, which is used to adjust the contribution proportion of the cleaning effect to the overall score, and the cleaning effect is the core target for evaluating the adaptation combination;

[0191] E clean represents the cleaning effect corresponding to the adaptation combination, which includes key indicators such as dirt removal rate and cleaning uniformity, and directly reflects whether the combination can meet the cleaning demand;

[0192] ξ represents the score coefficient of the scheduling priority, which is used to adjust the contribution proportion of the resource scheduling path priority to the overall score, and reasonable scheduling is an important support for ensuring stable implementation of the cleaning effect;

[0193] P dispThe resource scheduling path priority corresponding to the combination reflects the rationality and efficiency of the combination in the resource scheduling link;

[0194] The formula is used for calculating the effect score of the adaptive combination, and the core logic is: taking the corresponding cleaning effect (such as dirt removal rate) of the adaptive combination as the core, combining the corresponding resource scheduling path priority (the higher the priority, the more reasonable the scheduling, and the stronger the positive support to the effect), adjusting the contribution of the two to the effect score through the scoring coefficient, generating a quantitative effect score, providing a basis for identifying the optimal adaptive combination and optimizing the subsequent task allocation strategy, and avoiding one-sided evaluation of only focusing on cleaning effect or only focusing on scheduling;

[0195] The effect score calculated by the formula can evaluate the adaptive combination from the "cleaning effect + scheduling rationality" two-dimensional dimension, solve the optimization deviation problem caused by traditional evaluation only focusing on a single dimension, help accurately identify the adaptive combination with better comprehensive performance, and feed it back to the task allocation link, realize the continuous iterative optimization of the task allocation strategy, make the subsequent cleaning operation not only achieve the expected cleaning effect, but also reduce resource waste through reasonable scheduling, improve the quality and efficiency of the whole cleaning operation, and form a closed loop of "evaluation-optimization-application";

[0196] S46. Strategy optimization sub-step: feed the optimal adaptive combination to the S2 intelligent task allocation step as a reference basis for subsequent task allocation, and continuously optimize the path planning and equipment matching accuracy.

[0197] In the embodiments of the present application, the S3 dynamic coordination control step further comprises:

[0198] S313. Extreme environment triggering sub-step: when S14 detects an extreme environment (wind speed ≥ preset limit value, short-term heavy rain), immediately trigger path emergency adjustment;

[0199] S314. Path optimization sub-step: based on the extreme environment data, optimize the flight path of all drones in the region, including shortening the flight distance, reducing the flight height, avoiding the windward area, and temporarily suspending high-risk cleaning actions;

[0200] The flight height adjustment value under the extreme environment is calculated by the following algorithm formula:

[0201] ΔH=v×S adapt ·(H std -H safe );

[0202] Wherein, ΔH represents the flight height adjustment value under the extreme environment, a negative value indicates that the flight height needs to be reduced, and a positive value indicates that the flight height needs to be increased (usually negative under extreme environment, to reduce flight risk), used to determine the specific amplitude of the height adjustment of the drone;

[0203] v represents a height adjustment coefficient, used to control the influence range of the historical optimal adaptation combination effect score and the height difference value on the adjustment value, to avoid excessive or insufficient adjustment range;

[0204] S adapt represents the effect score of the adaptation combination, and a higher score indicates that the flight parameters (such as height) of the combination are more matched with the characteristics of the region, which can be used as a reference benchmark for adjustment in extreme environments;

[0205] H std represents the regular flight height of the unmanned aerial vehicle, i.e., the reasonable flight height of the region in non-extreme environments;

[0206] H safe represents the safe flight height in extreme environments, i.e., the height threshold at which the unmanned aerial vehicle can safely fly in extreme environments (such as strong winds and heavy rain), which is usually lower than the regular flight height;

[0207] The formula is used to calculate the flight height adjustment value in extreme environments, and the core logic is: based on the effect score of the historical adaptation combination of the region (a higher score indicates that the flight parameters of the combination are more adapted to the characteristics of the region, and the reference is stronger), combined with the difference between the regular flight height of the unmanned aerial vehicle and the safe flight height in extreme environments (the larger the difference, the larger the basis for adjustment), the adjustment range is controlled through the height adjustment coefficient, to generate a quantitative height adjustment value, to provide a basis for path optimization in extreme environments, and to avoid safety risks or reduced cleaning effect caused by improper height adjustment;

[0208] The flight height adjustment value calculated by the formula can combine the historical optimal adaptation experience of the region and the safety requirements of extreme environments, to realize precise height adjustment and avoid safety risks caused by relying on experience judgment for height adjustment in traditional extreme environments (such as remaining at a high-risk height due to insufficient adjustment);

[0209] S315. Safety scheduling sub-step: if the duration of the extreme environment is greater than or equal to the preset time length, the unmanned aerial vehicles in the region are dispatched to a preset safe area, the task is suspended, and a restart plan after the environment recovers is generated, and after the environment recovers, the task path is restarted or adjusted based on the dirt residue data and the device state.

[0210] The embodiments of the present application are disclosed, but are not limited thereto, and those skilled in the art can easily understand the spirit of the present application and make different inferences and changes based on the above embodiments, as long as they do not deviate from the spirit of the present application, and are within the protection scope of the present application.

Claims

1. A dynamic planning method for unmanned aerial vehicle (UAV) photovoltaic cleaning paths based on dirt distribution, characterized in that, Includes the following steps: S1. Distributed data acquisition steps: Simultaneously collect data on the surface dirt characteristics of photovoltaic panels, the functional adaptability data of drone cleaning equipment, the terrain data of photovoltaic panel arrays, and the energy consumption characteristics of drones. The dirt characteristics data includes dirt adhesion, chemical properties, and regional distribution density. The equipment functional adaptability data includes the cleaning components' ability to handle different types of dirt and their adaptability to cross-regional operations. The energy consumption characteristics data includes energy consumption differences under different dirt cleaning scenarios and different terrains. All data are classified according to the "dirt-equipment-energy consumption-region" correlation dimension. S2. Intelligent task allocation steps: Based on the associated data classified in S1, a four-dimensional dynamic decision matrix of "dirt characteristics adaptability - equipment function matching degree - energy consumption cost-effectiveness - regional adaptability" is constructed. An improved ant colony algorithm that integrates "equipment capability pre-assessment + regional adaptability prediction" is adopted to coordinate the allocation of cleaning tasks and paths for each drone. S3. Dynamic Collaborative Control Steps: Achieve collision-free collaboration among multiple drones. Through a three-feedback mechanism of "cleaning effect - equipment status - regional environment", the dirty characteristic data in S1 and the task path and equipment adaptation strategy in S2 are corrected in real time. At the same time, the "resource pre-scheduling + task splitting + regional collaboration" mode is used to handle sudden dirty events.

2. The method for dynamic planning of UAV photovoltaic cleaning path based on dirt distribution according to claim 1, characterized in that, The S1 distributed data acquisition step specifically includes the following sub-steps: S11. Dirt Characteristics Acquisition Sub-step: By integrating multispectral sensors, pressure sensing probes and image recognition units into each UAV, simultaneously acquire data on dirt type, coverage area, adhesion, chemical properties and regional distribution density, and generate a local dirt heat map with "characteristic-region" dual labels; S12. Equipment Function Adaptability Acquisition Sub-step: By integrating the sensing units into each UAV, monitor in real time the cleaning components' ability to handle different dirt characteristics and the UAV's power system's adaptability to different terrains, and generate an equipment "function-area" adaptability list. S13. Energy consumption characteristic acquisition sub-step: Based on the pre-stored photovoltaic panel array terrain information, combined with the real-time collected UAV "cleaning action-terrain-energy consumption" correspondence data, generate energy consumption characteristic heat maps under different cleaning scenarios and different terrains; S14. Regional Environmental Data Acquisition Sub-step: Real-time data on wind speed and humidity in each region is collected using fixed sensors deployed at the photovoltaic power station. This data is used to correct the energy consumption data in S13 and subsequent path stability parameters.

3. The method for dynamic planning of UAV photovoltaic cleaning path based on dirt distribution according to claim 1, characterized in that, The improvement to the traditional ant colony algorithm in the S2 intelligent task allocation step specifically includes the following sub-steps: S21. Dual Adaptation Pre-evaluation Sub-step: Based on the device "function-area" adaptability list generated in S12, select drones whose matching degree with the target area's terrain and dirt characteristics is ≥ the preset threshold, and exclude drones with insufficient adaptability before proceeding to the path planning stage. S22. Heuristic Factor Weight Adjustment Sub-step: Set "Dirty Characteristics Adaptability" and "Regional Adaptability" as core heuristic factors, and set "Equipment Function Matching Degree" and "Energy Consumption Cost-Effectiveness" as adjustment factors, dynamically adjusting the weight of each factor: Increase the weight of "equipment cleaning intensity matching" for areas with high adhesion and dirt. Increase the weight of "drone terrain adaptability" in steep slope areas; Increase the weight of "equipment waterproof performance matching degree" for high humidity areas; The dynamic weights of each heuristic factor are calculated using the following algorithm formula: Among them, W i This represents the dynamic weight of the i-th heuristic factor, used to quantify the importance of the heuristic factor in task allocation decisions. The larger the weight value, the stronger its guiding effect on the direction of task allocation. α represents the weighting coefficient of fitness, which is used to adjust the contribution ratio of fitness corresponding to the heuristic factor to the dynamic weight. It can be flexibly adjusted according to the priority requirements of different cleaning scenarios. F i It represents the fit degree corresponding to the i-th heuristic factor, covering the fit degree of dirt characteristics, regional fit degree, and equipment function fit degree. It is a quantitative indicator for measuring the degree of matching between the UAV and the mission requirements. β represents the weighting coefficient of energy consumption characteristics, which is used to adjust the contribution ratio of energy consumption characteristics to dynamic weights, and together with α, achieves a balance between adaptability and energy economy. E base This is the basic value of the energy consumption characteristics of the drone in the target area, and its reciprocal reflects the energy economy. S23. Dual-mechanism prediction sub-step: Based on historical cleaning data, current equipment status and regional environmental data, predict the probability, effect and equipment wear of each drone to complete the task, and prioritize the allocation of tasks to drones with "high matching degree + high expected effect + low wear".

4. The method for dynamic planning of UAV photovoltaic cleaning path based on dirt distribution according to claim 1, characterized in that, The collision-free coordination in the S3 dynamic cooperative control step specifically includes the following sub-steps: S31. Complementary obstacle avoidance sub-step: Based on the real-time location data, equipment function status and regional environmental data of each UAV, the logic of "speed field obstacle avoidance + function-regional complementary obstacle avoidance" is adopted. When the paths of two UAVs intersect, the path of the "high adhesion dirt-specific model" or "steep slope area specific model" is prioritized, and the other party adjusts its path and flight speed in combination with wind speed data. The path adjustment amount for the drone is calculated using the following algorithm formula: ΔP=k×W func-area ·(V max -V curr ); Wherein, ΔP represents the path adjustment amount of the drone, which can be reflected as lateral offset distance and altitude adjustment value, and is used to quantify the path parameters that the drone needs to adjust in order to avoid other drones; k represents the path adjustment coefficient, which is used to control the overall magnitude of the path adjustment, avoiding excessive adjustment that could cause task delays or insufficient adjustment that could not achieve effective obstacle avoidance. W func-area The dynamic weights corresponding to the "function-region" adaptability are as follows: the higher the weight value, the stronger the adaptability of the UAV to the current region's mission, and the more significant the path adjustments that other UAVs need to make for it. V max The maximum safe flight speed of a drone is the upper limit of the safe flight speed that a drone can achieve in the current environment, which is determined by both the drone's performance and the regional environment. V curr This indicates the drone's current flight speed, i.e., the actual flight speed of the drone when the path crosses. The difference between this and the maximum safe flight speed reflects the drone's speed adjustment potential. S32. Fault Emergency Handover Sub-step: When a certain UAV triggers an equipment fault warning, immediately search for nearby "function-area dual-adaptation" replacement UAVs, split and hand over the unfinished tasks of the faulty UAV according to "dirt characteristics-area", and ensure that the tasks are continuous and the matching degree remains unchanged. S33. Wind speed adaptation adjustment sub-step: For areas where the wind speed is greater than or equal to the preset threshold, dynamically adjust the drone's flight altitude and path spacing within the area to increase obstacle avoidance redundancy.

5. The method for dynamic planning of UAV photovoltaic cleaning path based on dirt distribution according to claim 4, characterized in that, The three-feedback correction in the S3 dynamic cooperative control step specifically includes the following sub-steps: S34. Multi-data receiving sub-step: Receive "post-cleaning image data + equipment status change data + real-time area environment data" after each UAV has completed cleaning; S35. Cause Determination and Correction Sub-step: Analyze the amount of residual dirt through image data, and determine the cause of the problem by combining equipment status data and regional environmental data. If the amount of dirt residue exceeds the standard, the equipment is normal, and the environment is normal, adjust the path density in that area. If the amount of dirt residue exceeds the standard, the equipment is abnormal, and the environment is normal, the "functionally adapted" drone will be reassigned. If the amount of dirt residue meets the standard, the equipment is malfunctioning, and the humidity exceeds the standard, update the equipment's "Function-Area" compatibility list; The path density correction factor is calculated using the following algorithm formula: Among them, K d This represents the path density correction factor, which is used to quantify the proportion of the original path density that needs to be adjusted. A factor greater than 1 indicates that the path density needs to be increased, while a factor close to 1 indicates that the original path density is basically reasonable. γ represents the correction coefficient adjustment factor, which is used to control the degree of influence of the path adjustment amount on the correction coefficient, and avoid over-correction due to fluctuations in a single adjustment amount; ΔP is the drone path adjustment amount, and its absolute value reflects the degree of deviation between the original path plan and the actual scenario (such as obstacle avoidance requirements, cleaning requirements); P std The standard path spacing for the target area, which is a reasonable path spacing preset based on the dirt characteristics and terrain conditions of the area, is the benchmark for judging whether the path density is reasonable. S36. Data update sub-step: Based on the judgment result of S35, update the "characteristic-region" dual label of the global dirt heat map, the energy consumption characteristic heat map, and the equipment "function-region" compatibility list in real time.

6. The method for dynamic planning of UAV photovoltaic cleaning path based on dirt distribution according to claim 5, characterized in that, The handling of sudden dirt events in the S3 dynamic collaborative control step specifically includes the following sub-steps: S37. Emergency Level Triggering Sub-Step: Real-time monitoring of the concentration and characteristics of dirt on the photovoltaic panel surface and regional environmental data. When a local area experiences a sudden increase in dirt concentration, special characteristics, or is accompanied by extreme weather, the corresponding level of emergency response is triggered, and the "Emergency Dirt Characteristics - Emergency Area Environment" label is marked. S38. Priority scheduling sub-step: From the currently working drones, select drones that are "functionally adapted to emergency dirty characteristics + regionally adapted to emergency environment + with few remaining tasks", transfer their unfinished ordinary tasks to nearby adapted drones according to "region", and then schedule the drone to perform the emergency task. S39. Phased processing sub-steps: If there are no suitable drones available for immediate dispatch, initiate "off-site suitable drone pre-dispatch" and break down the emergency area task into "temporary protective cleaning + deep precision cleaning". First, nearby ordinary drones perform temporary protective cleaning, and then perform deep precision cleaning after the suitable drones arrive.

7. The method for dynamic planning of UAV photovoltaic cleaning path based on dirt distribution according to claim 6, characterized in that, The S3 dynamic collaborative control steps also include: S310. Loss Prediction Sub-step: Based on real-time equipment status data, historical loss data and current task characteristics, establish an equipment loss prediction model and calculate the estimated loss value of each UAV after completing the current task. S311. Task handover warning sub-step: When the estimated loss value of a certain UAV is greater than or equal to the preset safety threshold, the "task handover warning" is triggered, the surrounding compatible UAVs are searched and the unfinished tasks are handed over step by step, and an equipment maintenance reminder is generated at the same time. S312. Loss Balancing Sub-step: For UAVs that have been handling high-loss tasks for a long time, adjust their subsequent task allocation strategy, increase the proportion of low-loss tasks, and achieve equipment loss balancing.

8. The method for dynamic planning of UAV photovoltaic cleaning path based on dirt distribution according to claim 7, characterized in that, The S3 dynamic collaborative control steps also include: S313. Extreme Environment Triggering Sub-step: When S14 detects an extreme environment, it immediately triggers an emergency path adjustment; S314. Path optimization sub-step: Based on extreme environment data, optimize the flight paths of all drones in the area, including shortening the flight distance, reducing the flight altitude, avoiding the windward area, and suspending high-risk cleaning operations. S315. Safety scheduling sub-step: If the duration of the extreme environment is greater than or equal to the preset duration, the drone in the area is scheduled to the preset safe area, the mission is paused and a restart plan is generated after the environment is restored. After the environment is restored, the mission path is restarted or adjusted based on the dirt residue data and equipment status.

9. The method for dynamic planning of UAV photovoltaic cleaning path based on dirt distribution according to claim 1, characterized in that, The method further includes: S4. Visualized monitoring and intervention steps: Construct a digital twin interface that links "dirt-equipment-path-area" to display the matching status of each drone with the dirty scene and regional environment, the task iteration process and energy consumption optimization effect in real time, and support operation and maintenance personnel to make precise intervention and cross-regional resource scheduling based on the associated data; The S4 visualization monitoring and intervention steps also include: S41. Resource Status Identification Sub-step: Based on the correlation data of "dirt density - number of devices - device compatibility" of each area displayed in the digital twin interface, identify areas with idle resources and areas with scarce resources; S42. Scheduling instruction execution sub-step: Operation and maintenance personnel initiate cross-regional resource scheduling instructions through the visual interface. Based on the principle of "device-region dual adaptation", the system plans the cross-regional scheduling path of idle equipment and adjusts the task allocation scheme of the target region at the same time. S43. Dispatch route optimization sub-step: During cross-regional dispatching, the dispatch route is dynamically optimized by combining environmental data of the areas along the route to ensure that the equipment arrives safely and efficiently.

10. The method for dynamic planning of UAV photovoltaic cleaning path based on dirt distribution according to claim 9, characterized in that, The S4 visualization monitoring and intervention steps also include: S44. Data storage sub-step: Store the "initial dirt data - equipment information - cleaning path - post-cleaning data - equipment wear data" for each area during each cleaning, forming a cleaning effect traceability database; S45. Correlation Analysis Sub-step: Based on the traceability database data, periodically generate a correlation analysis report on "dirt characteristics - equipment compatibility - cleaning effect" to identify the optimal compatibility combination; S46. Strategy Optimization Sub-step: Feedback the optimal matching combination to the S2 intelligent task allocation step as a reference for subsequent task allocation, and continuously optimize path planning and device matching accuracy.

Images