Method, device and storage medium for cleaning a drone

Abstract

This application discloses a control method, device, and storage medium for a cleaning drone. The method relates to the field of drone control technology. The method includes: acquiring the echo signal of a ground ranging radar; adjusting the signal-to-noise ratio (SNR) threshold and radar cross section (RCS) threshold according to the water mist concentration of the current cleaning operation; extracting target points in the echo signal that simultaneously satisfy both the SNR and RCS thresholds; outputting the current distance value of the target points to filter out water mist clutter interference; the current distance value representing the distance between the cleaning drone and a target object below its horizontal plane; comparing the current distance value with a preset terrain-following distance; and controlling the cleaning drone to ascend when the current distance value is less than the preset terrain-following distance to approach the preset distance. This application improves the operational safety and flight stability of cleaning drones in photovoltaic array scenarios.

Description

Technical Field

[0001] This application relates to the field of unmanned aerial vehicle (UAV) control technology, and in particular to a control method, device and storage medium for cleaning UAVs. Background Technology

[0002] With the expansion of photovoltaic power plants and the upgrading of operation and maintenance automation requirements, using drones to clean photovoltaic arrays has become an important means to ensure power generation efficiency. These cleaning drones need to fly above the photovoltaic modules at a fixed ground-mimicking distance, controlling their flight altitude by sensing the distance below in real time, ensuring that they are neither too high and affect the cleaning effect, nor too low and collide with the panels.

[0003] Currently, terrain-following flight is typically achieved using ground-range radar. The radar continuously emits electromagnetic waves downwards and receives reflected signals, calculating the distance between the UAV and targets below. The flight control system then executes closed-loop altitude control based on this calculation. This method can operate reliably on continuously flat photovoltaic panel surfaces.

[0004] However, photovoltaic power plants consist of a large number of discrete components arranged in arrays, with gaps of about 1 meter wide commonly present. These gaps are unobstructed by photovoltaic panels, allowing radar electromagnetic waves to penetrate directly to the ground or the structure beneath the panels. This causes a sudden jump in the distance detected by ground-based radar when a drone flies over these gaps, abruptly increasing from the normal 1-2 meters to several meters or even tens of meters. Conventional terrain-following control logic interprets this jump as excessive altitude and immediately outputs a descent command, causing the drone to rapidly descend to approach the set terrain-following distance. Because the time spent crossing the gaps is extremely short, the drone begins to descend before fully crossing them, making it highly susceptible to collisions with the next row of photovoltaic modules behind the gap, resulting in equipment damage. Summary of the Invention

[0005] The main objective of this application is to provide a control method, device, and storage medium for cleaning drones, aiming to solve the technical problem of abnormal drone descent caused by gaps in photovoltaic panels.

[0006] To achieve the above objectives, this application provides a control method for a cleaning drone, applied to a cleaning drone equipped with a ground ranging radar, and the control method for the cleaning drone includes: Acquire the echo signal of the ground ranging radar; The signal-to-noise ratio (SNR) threshold and radar cross section (RCS) threshold are adjusted according to the water mist concentration of the current cleaning operation. Target points that simultaneously satisfy the SNR and RCS thresholds are extracted from the echo signal, and the current distance value of the target points is output to filter out water mist clutter interference. The current distance value represents the distance between the cleaning drone and the target object below its horizontal plane. The real-time attitude angles of the cleaning drone are obtained from the inertial measurement unit of the flight control system. The real-time attitude angles include pitch angle and roll angle. The combined tilt angle of the cleaning drone is calculated based on the pitch angle, the roll angle, and the first preset formula; Based on the synthesized tilt angle and the second preset formula, the current distance value is used as the slant distance to calculate the updated current distance value; The updated current distance value is compared with the preset terrain-following distance. When the current distance value is less than the terrain-following distance, the cleaning drone is controlled to ascend to approach the terrain-following distance.

[0007] In one embodiment, adjusting the signal-to-noise ratio (SNR) threshold and radar cross section (RCS) threshold based on the water mist concentration of the current cleaning operation includes: The water mist concentration is determined based on the air humidity of the current working environment, the echo signal strength of the ground ranging radar, and the output power of the water pump in the current cleaning operation. When the water mist concentration exceeds a preset concentration threshold, the SNR threshold and / or the RCS threshold are increased to enhance the filtering capability of water mist clutter. When the water mist concentration is lower than or equal to the preset concentration threshold, the SNR threshold and / or the RCS threshold are reduced to the default value to improve the detection sensitivity of photovoltaic panel targets.

[0008] In one embodiment, determining the water mist concentration based on the air humidity of the current working environment, the echo signal strength of the ground ranging radar, and the output power of the water pump in the current cleaning operation includes: The system obtains the air humidity collected by the humidity sensor on the cleaning drone and the water pump output power of the cleaning drone during the current cleaning operation. The echo signal attenuation is calculated based on the reference echo signal strength of the ground ranging radar when the nozzle is not turned on and the echo signal strength after the nozzle is turned on. The water mist concentration is calculated based on the air humidity, the water pump output power, and the echo signal attenuation, using a water mist evaluation formula. The water mist evaluation formula includes: Cmist=α*Hcurr+β*Ppump+γ*ΔSCmist Where α is the humidity weighting coefficient, β is the power weighting coefficient, γ is the attenuation weighting coefficient, Hcurr is the air humidity, Ppump is the water pump output power, ΔSCmist is the echo signal attenuation, and Cmist is the water mist concentration.

[0009] In one embodiment, the first preset formula is: θtilt=arccos(cosθpitch*cosθroll); Where θtilt is the composite tilt angle, θpitch is the pitch angle, and θroll is the roll angle; The second preset formula is: Htrue = Rslant * cos * θtilt; Where Htrue is the updated current distance value, and Rslant is the current distance value.

[0010] In one embodiment, after calculating the combined tilt angle of the cleaning drone based on the pitch angle, the roll angle, and a first preset formula, the process includes: When the absolute value of the synthesized tilt angle exceeds the preset tilt threshold, the SNR threshold and / or the RCS threshold are increased to reduce the interference of water mist density distribution changes on ranging accuracy under crosswind conditions.

[0011] In one embodiment, after calculating the combined tilt angle of the cleaning drone based on the pitch angle, the roll angle, and a first preset formula, the process includes: When the synthesized tilt angle exceeds the preset limit safety threshold, it is determined that the current flight attitude has exceeded the safety compensation range of terrain-following flight. Control the cleaning drone to perform hovering or emergency landing, and suspend the current cleaning operation.

[0012] In one embodiment, comparing the updated current distance value with a preset terrain simulation distance includes: Obtain pre-stored flight path information, wherein the positions of the photovoltaic array gaps are marked in the flight path information; When the cleaning drone is determined to be above the gap based on the flight path information, the nozzle is controlled to close. When the cleaning drone is determined to have crossed the gap and is now above the photovoltaic panel again based on the flight path information, the control nozzle is restarted to resume the current cleaning operation.

[0013] In addition, to achieve the above objectives, this application also provides a control device for a cleaning drone, the control device for the cleaning drone comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the control method for the cleaning drone as described above.

[0014] In addition, to achieve the above objectives, this application also provides a storage medium, which is a computer-readable storage medium, on which a program for implementing a control method for a cleaning drone is stored, and the program for implementing the control method for a cleaning drone is executed by a processor to implement the steps of the control method for a cleaning drone as described above.

[0015] This application provides a control method for a cleaning drone. First, it acquires the echo signal from a ground-based ranging radar. Based on the water mist concentration during the current cleaning operation, it adjusts the signal-to-noise ratio (SNR) threshold and the radar cross section (RCS) threshold. It then extracts target points that simultaneously meet both the SNR and RCS thresholds and outputs the current distance value after filtering out water mist clutter interference, accurately representing the distance between the cleaning drone and the target below. After obtaining the accurate current distance value, it compares this value with a preset ground-following distance to determine if the cleaning drone faces an abnormal descent risk. When the current distance value is detected to be less than the ground-following distance, it indicates the presence of an obstacle or photovoltaic panel below the cleaning drone. The drone is then controlled to ascend to approach the ground-following distance, ensuring operational safety. Conversely, when the current distance value is detected to be greater than or equal to the ground-following distance, it indicates that there may be a gap in the photovoltaic array below the cleaning drone. The drone is then controlled to maintain its current altitude and is prohibited from descent, thus preventing the cleaning drone from misjudging its altitude due to sudden changes in distance when flying over photovoltaic array gaps, causing abnormal descent and collisions with the rear photovoltaic modules.

[0016] In summary, this application overcomes the technical defects of traditional ground-following flight control methods that cause abnormal descent due to ranging jumps when flying over gaps in photovoltaic arrays by using the accurate distance value after filtering out water mist interference as the judgment basis and combining it with an asymmetric ground-following flight control strategy. This avoids the risk of collision between the cleaning drone and the rear photovoltaic modules and improves the operational safety and flight stability of the cleaning drone in photovoltaic array scenarios. Attached Figure Description

[0017] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0018] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a flowchart illustrating an embodiment of the control method for cleaning drones according to this application. Figure 2This is a flowchart illustrating Embodiment 4 of the control method for cleaning drones according to this application. Figure 3 This is a schematic diagram of the hardware architecture involved in the control equipment of the cleaning drone in this application.

[0020] The purpose, features, and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0021] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of this application and are not intended to limit this application.

[0022] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0023] Currently, photovoltaic power plants consist of a large number of discrete modules arranged in arrays, with gaps of about 1 meter wide commonly present. These gaps are unobstructed by photovoltaic panels, allowing radar electromagnetic waves to penetrate directly to the ground or the structure beneath the panels. This causes a sudden jump in the distance detected by ground-based radar when a drone flies over these gaps, abruptly increasing from the normal 1-2 meters to several meters or even tens of meters. Conventional terrain-following control logic interprets this distance jump as excessive altitude and immediately outputs a descent command, causing the drone to rapidly descend to approach the set terrain-following distance. Because the time spent crossing the gaps is extremely short, the drone begins to descend before fully crossing the gap, making it highly susceptible to collisions with the next row of photovoltaic modules behind the gap, resulting in equipment damage.

[0024] The main solution of this application is as follows: acquire the echo signal of the ground ranging radar; adjust the signal-to-noise ratio (SNR) threshold and radar cross section (RCS) threshold according to the water mist concentration of the current cleaning operation, and extract the target points in the echo signal that simultaneously meet the SNR threshold and the RCS threshold, and output the current distance value of the target points to filter out water mist clutter interference. The current distance value represents the distance between the cleaning drone and the target object below its horizontal plane; compare the current distance value with a preset terrain-following distance, and when the current distance value is less than the terrain-following distance, control the cleaning drone to ascend to approach the terrain-following distance.

[0025] This application overcomes the technical defects of traditional ground-following flight control methods that cause abnormal descent due to ranging jumps when flying over gaps in photovoltaic arrays by using the accurate distance value after filtering out water mist interference as the judgment basis and combining it with an asymmetric ground-following flight control strategy. It avoids the risk of collision between the cleaning drone and the rear photovoltaic modules and improves the operational safety and flight stability of the cleaning drone in photovoltaic array scenarios.

[0026] It should be noted that the executing entity in this embodiment can be the control system of a cleaning drone, or a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, or mobile phone, or a control device or cleaning drone capable of performing the above functions. This embodiment does not specifically limit the specific implementation. The following uses a cleaning drone as the executing entity to describe this embodiment and the following embodiments.

[0027] Based on this, Embodiment 1 of this application proposes a control method for a cleaning drone. Please refer to... Figure 1 The control method for the cleaning drone includes steps S10 to S30: Step S10: Obtain the echo signal of the ground ranging radar.

[0028] In this embodiment, the ground ranging radar refers to a millimeter-wave radar installed below the cleaning drone with its beam pointing towards the ground, and its transmitted signal is a linear frequency modulated continuous wave. The echo signal refers to the electromagnetic wave signal captured by the radar receiving antenna and reflected back from the target object below, containing amplitude, phase, and frequency difference information, which can reflect the distance and scattering characteristics of the target object.

[0029] As an optional implementation, the flight control system sends a data request command to the ground ranging radar at a fixed frequency via a serial communication interface. Upon receiving the request, the ground ranging radar digitizes the raw intermediate frequency signal acquired during the current frequency modulation cycle and transmits it back to the flight control system in the form of a data packet via the same serial interface. After parsing the data packet, the flight control system obtains an echo signal matrix containing complex sample values ​​of multiple range cells. The rows of this matrix correspond to the frequency modulation cycle number, the columns correspond to each range cell, and the matrix elements are complex values ​​composed of in-phase and quadrature components, which are directly used for subsequent range-dimensional fast Fourier transform and target detection processing.

[0030] As an alternative implementation, the ground ranging radar performs intermediate frequency signal sampling and range-dimensional fast Fourier transform within its own digital signal processor. The resulting range-amplitude spectrum is then used as an echo signal and continuously pushed to the flight control system in binary frame format via a serial port. Each echo signal frame contains a sequence of amplitude values ​​arranged sequentially by range cell index and the corresponding range cell number. The flight control system does not need to perform a transformation operation and can directly perform subsequent target detection based on this amplitude sequence.

[0031] Step S20: Adjust the signal-to-noise ratio (SNR) threshold and radar cross section (RCS) threshold according to the water mist concentration of the current cleaning operation, extract the target points in the echo signal that simultaneously meet the SNR threshold and the RCS threshold, and output the current distance value of the target points to filter out water mist clutter interference. The current distance value represents the distance between the cleaning drone and the target object below its horizontal plane.

[0032] In this embodiment, the signal-to-noise ratio (SNR) threshold refers to the minimum SNR threshold value used to determine whether a target point has sufficient signal strength, and the radar cross section (RCS) threshold refers to the minimum radar cross section threshold value used to determine whether a target point has sufficient physical scale. Water mist clutter interference refers to the low-intensity, low-RCS, spatially diffuse false echo signals formed by water mist particles ejected from a high-pressure nozzle in the radar beam. In the range spectrum, it manifests as a dense but low-amplitude noise floor elevation.

[0033] As an optional implementation, the air humidity value in the current cleaning environment, the reference echo base amplitude of the ground ranging radar with the nozzle closed, and the current water pump output power are first acquired. A water mist concentration assessment value is calculated based on the air humidity value, water pump output power, and reference echo base amplitude. Specifically, the air humidity value is multiplied by a preset humidity weight, the water pump output power is multiplied by a preset power weight, and the change in the reference echo base amplitude relative to the unsprayed state is multiplied by a preset attenuation weight, and then summed to obtain the water mist concentration assessment value. When the water mist concentration assessment value exceeds a preset level threshold, the SNR threshold is increased from the default value to the first highest threshold, and the RCS threshold is increased from the default value to the second highest threshold to enhance the filtering effect of water mist clutter; when the water mist concentration assessment value does not exceed the preset level threshold, the SNR threshold and RCS threshold are maintained at their respective default values. After the threshold is determined, the signal-to-noise ratio (SNR) and radar cross section (RCS) of all candidate points obtained by constant false alarm rate (CFAR) detection of the echo signal are calculated one by one. Only candidate points that simultaneously satisfy the SNR threshold and RCS threshold are retained as valid target points. The distance value corresponding to the range cell index in the valid target points is used as the current distance value and output to the flight control module.

[0034] As an alternative implementation, instead of explicitly calculating the water mist concentration assessment value, clutter statistics are performed on the range-amplitude spectrum of the echo signal, continuously updating the mean and standard deviation of background noise at each range cell. When multiple consecutive range cells within a preset short-range segment are detected to have amplitude values ​​significantly higher than the background noise but lower than the typical photovoltaic panel echo amplitude, water mist clutter is determined to exist in that short-range segment. At this time, the SNR threshold is switched from the default value to the mean background noise plus a first dynamic bias, and the RCS threshold is switched to the estimated upper limit based on the water mist echo scattering cross section plus a preset protection margin. The dynamically adjusted SNR and RCS thresholds are used to filter candidate points for constant false alarm detection, extract the true target points of the photovoltaic panel, and output their distance values. This method drives the adaptive adjustment of the threshold by directly analyzing the statistical characteristics of water mist clutter in the range spectrum, without requiring additional humidity sensors and water pump power signals.

[0035] Step S30: Compare the current distance value with the preset terrain simulation distance. When the current distance value is less than the preset terrain simulation distance, control the cleaning drone to rise to approach the preset terrain simulation distance.

[0036] In this embodiment, the terrain-following setting distance refers to the desired vertical working distance between the drone and the photovoltaic panel surface, preset according to the cleaning process requirements. Approaching means that the current distance value is gradually brought closer to the terrain-following setting distance by outputting ascent commands through the flight control system, rather than requiring instantaneous and precise attainment.

[0037] As an optional implementation, the flight control system compares the current distance value output in step S20 with the terrain-following preset distance in each control cycle. If the current distance value is less than the terrain-following preset distance, the distance difference is calculated and input into the cascaded proportional-integral-derivative controller to generate a vertical speed command. The vertical speed command is then calculated into the throttle increment of each motor to propel the UAV upward. During ascent, the current distance value and the terrain-following preset distance are continuously compared. When the absolute value of the difference falls within the preset allowable dead zone, ascent stops and the current altitude is maintained. If the current distance value is greater than or equal to the terrain-following preset distance, the flight control system does not generate any ascent command, maintains the current throttle output, and ensures that the UAV does not sink due to gaps below.

[0038] As an alternative implementation, the flight control system first sends the current distance value output in step S20 to a first-order low-pass filter to suppress high-frequency jitter in the distance value. The filtered distance value is then compared with the terrain-following preset distance. If the filtered distance value is less than the terrain-following preset distance, the ascent control sub-state machine is activated. This state machine gradually increases the desired altitude value in fixed steps, and the deviation between the desired altitude value and the filtered distance value drives the proportional controller to generate throttle increments until the filtered distance value returns to the dead zone of the terrain-following preset distance. If the filtered distance value remains above the terrain-following preset distance for several consecutive cycles, the state machine enters an altitude-holding state and resets the historical accumulation terms of the integral controller to eliminate interference from the integral saturation effect generated during past ascent processes on subsequent control.

[0039] For example, when a cleaning drone is performing a photovoltaic array cleaning task, the flight control system acquires echo signals from a ground ranging radar at a frequency of 50 Hz. Given the high air humidity and the water pump operating at full power, the control system determines that the water mist concentration is high and increases the signal-to-noise ratio (SNR) threshold from the default 12 dB to 15 dB, and the radar cross section (RCS) threshold from the default 0.01 square meters to 0.05 square meters. After constant false alarm rate (CFAR) detection, the echo signal generates seven candidate points. Five points with an SNR below 15 dB are discarded. Of the remaining two points, one with an RCS less than 0.05 square meters is identified as water mist clutter. Only one point meets both threshold conditions, corresponding to a distance of 1.48 meters. This distance is compared to the terrain-following set distance of 1.5 meters, and the result is less. The flight control system calculates the difference as 0.02 meters and outputs an ascent command. After the drone fine-tunes its altitude, the distance stabilizes around 1.5 meters. When the drone flies over the gaps in the photovoltaic array, the radar receives ground echoes from a greater distance due to the electromagnetic waves penetrating the gaps. However, these echoes still meet the signal-to-noise ratio and radar cross-section threshold, and are output as the current distance value, for example, 3.2 meters. This distance value is greater than the ground-following preset distance, so the flight control system does not generate ascent commands and prohibits descent. The drone maintains its current altitude and flies smoothly over the gaps. Subsequently, the distance value returns to 1.5 meters above the photovoltaic panel surface, without any sudden altitude change.

[0040] This embodiment incorporates water mist concentration into a threshold adjustment mechanism to adaptively distinguish between the true echo from the photovoltaic panel and water mist clutter, thereby accurately extracting the distance value representing the target object below in a water mist-filled cleaning environment. Based on this, an asymmetric control strategy is adopted that only executes ascent when the distance value is less than a set value. This prevents the UAV from sinking due to a sudden increase in distance when flying over gaps in the photovoltaic array, eliminating the safety hazard of collisions with rear-row components caused by misjudged altitude, and achieving stability and safety during ground-following flight in cleaning operations.

[0041] Based on any of the above embodiments, in Embodiment 2 of this application, adjusting the signal-to-noise ratio (SNR) threshold and radar cross section (RCS) threshold according to the water mist concentration of the current cleaning operation includes: Step S21: Determine the water mist concentration based on the air humidity of the current working environment, the echo signal strength of the ground ranging radar, and the output power of the water pump in the current cleaning operation.

[0042] In this embodiment, air humidity refers to the relative humidity value of the working environment collected by the humidity sensor on the cleaning drone. Echo signal strength refers to the average echo amplitude of the ground ranging radar on the near-range background range cells where no discrete target points are detected. Water pump output power refers to the current electrical power value of the high-pressure water pump in the cleaning module. Water mist concentration refers to a quantitative index characterizing the density of water mist in the space between the cleaning drone and the photovoltaic panel, obtained through a comprehensive evaluation of the above parameters.

[0043] As an optional implementation, the following steps are first taken: acquiring the relative humidity value output by the humidity sensor, the average echo amplitude value of the background range unit output by the ground ranging radar, and the real-time electrical power value fed back by the water pump drive circuit. The relative humidity value is multiplied by a preset first proportional coefficient to obtain the humidity contribution term; the real-time electrical power value is multiplied by a preset second proportional coefficient to obtain the power contribution term; the average echo amplitude value of the background range unit is subtracted from the reference echo amplitude value under the non-spraying calibration state to obtain the echo attenuation; and the echo attenuation is multiplied by a preset third proportional coefficient to obtain the attenuation contribution term. The sum of the humidity contribution term, power contribution term, and attenuation contribution term yields the water mist concentration value. This method achieves continuous quantitative assessment of water mist concentration by linearly weighting and fusing information from three dimensions.

[0044] As an alternative implementation, relative humidity, echo signal strength, and water pump output power are used as input features and input into a pre-constructed two-dimensional lookup table model. This model uses humidity and power levels as row and column indices, stores the water mist concentration levels calibrated under the corresponding operating conditions, and uses the current echo signal strength attenuation as a correction factor to adjust the lookup output, ultimately outputting a discretized water mist concentration level. This method pre-defines the mapping relationship of multi-dimensional parameters in the storage structure, requiring only a limited number of lookups and one correction operation during runtime, reducing the online computational overhead of the flight control system.

[0045] Step S22: When the water mist concentration exceeds a preset concentration threshold, increase the SNR threshold and / or the RCS threshold to enhance the filtering capability of water mist clutter.

[0046] In this embodiment, the preset concentration threshold refers to the critical water mist concentration value used to determine whether the water mist interference level needs to be strengthened for suppression. Increasing the SNR threshold means raising the signal-to-noise ratio threshold from the default operating value to a higher preset value, and increasing the RCS threshold means raising the radar cross section threshold from the default operating value to a higher preset value.

[0047] As an optional implementation, the water mist concentration value output in step S21 is compared with a preset concentration threshold. When the water mist concentration value is greater than the preset concentration threshold, the first-level SNR threshold and the first-level RCS threshold matching the current water mist concentration value are read from the parameter configuration table, and these two thresholds are written into the threshold register of the radar signal processing module, replacing the currently effective default SNR threshold and default RCS threshold. If the water mist concentration value further increases and exceeds the second preset concentration threshold, the second-level SNR threshold and the second-level RCS threshold are read and the register is updated. This method achieves threshold adjustment that matches the changes in water mist concentration in a step-like manner by dividing multiple concentration ranges and corresponding to multiple preset threshold levels.

[0048] As an alternative implementation, when the water mist concentration exceeds a preset concentration threshold, instead of using a pre-set backup switch, the water mist concentration is directly input into a preset threshold mapping function. This mapping function takes the water mist concentration as the independent variable and outputs the corresponding SNR and RCS thresholds. For example, when the water mist concentration is between the preset threshold and a first upper bound, the SNR threshold is a linearly increasing function of the default value and the concentration value, and the RCS threshold is a linearly increasing function of the default value and the square root of the concentration value. The flight control system calls this mapping function in each control cycle to calculate the real-time thresholds and update the radar signal processing parameters, ensuring that the threshold adjustment amplitude is continuously synchronized with the water mist concentration change trend.

[0049] Step S23: When the water mist concentration is lower than or equal to the preset concentration threshold, reduce the SNR threshold and / or the RCS threshold to the default value to improve the detection sensitivity of the photovoltaic panel target.

[0050] In this embodiment, reducing to the default value means restoring the SNR and RCS thresholds to the standard operating values ​​calibrated before the cleaning operation in a fog-free or low-fog environment. Detection sensitivity refers to the radar's ability to correctly identify photovoltaic panel targets from echo signals; the lower the threshold, the stronger the ability to acquire weak echo targets.

[0051] As an optional implementation, the water mist concentration value output in step S21 is continuously compared with a preset concentration threshold. When the water mist concentration value is detected to have fallen from above the preset concentration threshold to at or below the preset concentration threshold, a hysteresis timing judgment is initiated. After the water mist concentration value has not exceeded the preset concentration threshold for several consecutive control cycles, it is determined that the water mist state has stabilized and weakened. Subsequently, the values ​​in the SNR threshold register and the RCS threshold register are rewritten to the default SNR threshold and the default RCS threshold, respectively. This hysteresis mechanism avoids frequent switching between the default value and a high value due to instantaneous fluctuations in the water mist concentration near the threshold, preventing oscillations in the target detection state.

[0052] As an alternative implementation, when the water mist concentration falls below a preset concentration threshold, the threshold is not immediately reset. Instead, the current SNR and RCS thresholds are decreased periodically at a preset decreasing rate until the default values ​​are reached. During the decrease, the decrease step size for each control cycle is the difference between the default value and the high value divided by the preset number of recovery cycles. This method allows the threshold to smoothly transition from a high-intensity filtering state to a normal detection state, avoiding temporary missed detection of photovoltaic panel targets during the transition period due to sudden threshold changes.

[0053] For example, a cleaning drone initiates a cleaning operation in clear weather with an initial air humidity of 40%. The water pump operates at its rated power, and the background echo amplitude of the ground ranging radar is significantly attenuated due to the water mist. The control system calculates a water mist concentration of 78, exceeding the preset concentration threshold of 60. Therefore, it increases the SNR threshold from the default 12 dB to 16 dB and the RCS threshold from the default 0.01 square meters to 0.04 square meters. Subsequently, as the operation enters the edge area of ​​the photovoltaic array, the water pump power decreases due to the partial shutdown of the nozzles, and the air humidity decreases simultaneously, causing the water mist concentration to drop back to 45. After confirming for five consecutive control cycles that the water mist concentration does not exceed 60, the control system restores the SNR and RCS thresholds to their default values, and the radar regains its normal detection capability for weak echo targets on the photovoltaic panels.

[0054] This embodiment achieves real-time quantitative assessment of water mist concentration by comprehensively considering three operational status parameters: air humidity, radar echo signal strength, and water pump output power. The assessment results drive dynamic adjustments to the SNR and RCS thresholds, increasing when the water mist intensifies and decreasing when it weakens. This mechanism ensures that radar signal processing parameters are synchronized with the actual water mist conditions during the cleaning operation. It effectively filters out clutter interference when the water mist is dense and promptly restores high-sensitivity detection of photovoltaic panel targets after the water mist dissipates. This resolves the contradiction between filtration and detection capabilities under varying water mist conditions when using fixed thresholds.

[0055] Based on any of the above embodiments, in Embodiment 3 of this application, determining the water mist concentration according to the air humidity of the current working environment, the echo signal strength of the ground ranging radar, and the output power of the water pump in the current cleaning operation includes: Step S211: Obtain the air humidity collected by the humidity sensor on the cleaning drone, and obtain the water pump output power of the cleaning drone in the current cleaning operation.

[0056] In this embodiment, air humidity refers to the relative humidity percentage value output by the humidity sensor. Water pump output power refers to the real-time electrical power value fed back by the high-pressure water pump drive circuit in the cleaning module, measured in watts. The purpose of obtaining these two parameters is to provide a quantitative basis for the working environment and intensity of the subsequent water mist concentration assessment.

[0057] As an optional implementation, the flight control system reads the current value of the humidity sensor's data register at a fixed sampling period via its internal bus. Simultaneously, it sends a power query command to the water pump drive module via a serial communication interface. The water pump drive module returns the product of the current voltage and current as the water pump's output power. The flight control system stores the read air humidity value and water pump output power value in the water mist evaluation input buffer, along with a timestamp of the acquisition time, aligning it with the radar data timestamp for subsequent calculations.

[0058] As an alternative implementation, the airborne edge computing unit on the cleaning drone independently aggregates sensor data. It periodically receives humidity data frames reported by humidity sensor nodes and power data frames reported by the water pump controller via a controller area network bus in a broadcast manner. After parsing the data frames, it extracts the payload field to obtain the air humidity value and the water pump output power value. The edge computing unit responds to the arrival of data frames in an interrupt manner. When the timestamp deviation between the humidity data frame and the power data frame is within a preset tolerance window, they are paired and used as input parameters for the same evaluation cycle, written into shared memory for the water mist concentration calculation process to read.

[0059] Step S212: Calculate the echo signal attenuation based on the reference echo signal strength of the ground ranging radar when the nozzle is not turned on and the echo signal strength after the nozzle is turned on.

[0060] In this embodiment, the reference echo signal strength refers to the average background echo amplitude measured and recorded by the ground ranging radar at a designated reference distance unit before the cleaning operation begins and the nozzle is in the closed state. The echo signal strength after the nozzle is turned on refers to the real-time average echo amplitude at the same reference distance unit during the cleaning operation and when the nozzle is spraying water mist. The echo signal attenuation is the difference between the reference echo signal strength and the echo signal strength after the nozzle is turned on, used to quantify the degree of attenuation of radar electromagnetic waves by the water mist.

[0061] As an optional implementation, during the cleaning operation startup process, the flight control system first keeps the nozzle closed, records the echo amplitude values ​​of the ground ranging radar in a preset reference range cell over N consecutive frequency modulation cycles, calculates the arithmetic mean of the N amplitude values ​​as the reference echo signal strength, and stores it in non-volatile storage. After the cleaning operation begins and the nozzle is opened, the flight control system extracts the current echo amplitude value of the same reference range cell from the radar echo signal in each control cycle, subtracts it from the stored reference echo signal strength, and the difference is the echo signal attenuation amount for the current cycle. This attenuation amount changes in real time with the water mist concentration and is updated once in each control cycle.

[0062] As an alternative implementation, instead of pre-fixing the reference distance unit, the radar's range-amplitude spectrum is scanned across the entire range when the nozzle is off. The distance unit with the highest amplitude value and a distance value consistent with the nominal distance to the photovoltaic panel is selected as the reference unit, and its amplitude is recorded as the reference echo signal intensity. After the nozzle is turned on, the amplitude change of this distance unit is continuously tracked, and the difference between it and the reference echo signal intensity is used as the attenuation. If the amplitude of this distance unit decreases by more than a preset proportion due to water mist obstruction during the tracking process, a reference unit reselection mechanism is activated, and the unit with the second highest amplitude among the adjacent distance units is selected as the new reference unit, ensuring that the attenuation calculation is always based on the reliable distance unit where the actual echo from the photovoltaic panel is located.

[0063] Step S213: Calculate the water mist concentration based on the air humidity, the water pump output power, and the echo signal attenuation, using the water mist evaluation formula.

[0064] In this embodiment, the water mist evaluation formula is a mathematical expression that outputs the water mist concentration value with air humidity, water pump output power and echo signal attenuation as independent variables. The weight coefficients in the formula are determined through pre-calibration experiments.

[0065] As an optional implementation, the water mist evaluation formula adopts a linear weighted summation form, expressed as C = a*H + b*P + c*D, where C is the water mist concentration value, H is the air humidity, P is the water pump output power, D is the echo signal attenuation, a is the humidity weighting coefficient, b is the power weighting coefficient, and c is the attenuation weighting coefficient. The flight control system substitutes the air humidity value H obtained in step S211, the echo signal attenuation D obtained in step S212, and the water pump output power P obtained in step S211 into the above formula to calculate the current water mist concentration value C, which is used for subsequent comparison and judgment with a preset concentration threshold.

[0066] As an alternative implementation, the water mist evaluation formula adopts a nonlinear combination form, expressed as C = k1* H^alpha + k2 * P^beta + k3 * log(D+1), where k1, k2, and k3 are amplitude adjustment coefficients, alpha and beta are nonlinear exponents, and log is the natural logarithm. This formula describes the accelerating effect of air humidity on water mist generation, the saturating effect of water pump output power on water mist density, and the logarithmic compression characteristic of echo attenuation on water mist concentration through nonlinear mapping, giving the water mist concentration value a wider dynamic range under high water mist conditions. The flight control system calls this formula to complete the calculation in each evaluation cycle, and the obtained water mist concentration value is sent to the threshold comparison module.

[0067] Optionally, the water mist evaluation formula includes: Cmist=α*H curr +β*P pump +γ*ΔS Cmist Where α is the humidity weighting coefficient, β is the power weighting coefficient, γ is the attenuation weighting coefficient, and H curr For air humidity, P pump ΔS represents the output power of the water pump. Cmist Cmist represents the echo signal attenuation, and Cmist represents the water mist concentration.

[0068] For example, before the cleaning drone begins operation, the flight control system records a baseline echo signal strength of 85 dB / mW. After cleaning starts, the water pump operates at 300 watts, the humidity sensor measures an air humidity of 65 percent, and the echo signal strength at the same distance unit drops to 72 dB / mW, with an echo attenuation of 13 dB. The flight control system substitutes the air humidity of 65%, the water pump power of 300 watts, and the attenuation of 13 dB into the linear weighted formula C=0.5*H+0.3*P+0.8*D to calculate the water mist concentration value C=0.5*65+0.3*300+0.8*13=32.5+90+10.4=132.9. This water mist concentration value exceeds the preset concentration threshold of 100, and the control system immediately increases the SNR threshold from the default 12 dB to 15 dB and the RCS threshold from 0.01 square meters to 0.04 square meters, effectively filtering out water mist noise. Subsequently, the wind in the work area increased, the water mist was quickly dispersed, the air humidity dropped to 50%, the echo attenuation dropped to 3 dB, and the water pump power remained unchanged. The water mist concentration value C was recalculated as C = 0.5*50 + 0.3*300 + 0.8*3 = 25 + 90 + 2.4 = 117.4, which was still higher than the threshold, so the system continued to maintain a high threshold. When the water pump power dropped to 150 watts after completing one row of cleaning, the humidity further dropped to 40%, and the attenuation dropped to 1 dB. The calculated C = 0.5*40 + 0.3*150 + 0.8*1 = 20 + 45 + 0.8 = 65.8, which was lower than the threshold. The system restored the SNR threshold and RCS threshold to their default values, restoring high-sensitivity detection of the photovoltaic panel target.

[0069] This embodiment uses a method that integrates three parameters—air humidity, water pump output power, and radar echo attenuation—into a water mist evaluation formula. This establishes the determination of water mist concentration based on a mathematical foundation that quantitatively correlates the working environment, work intensity, and actual radar signal attenuation, achieving real-time, continuous, and precise quantification of water mist concentration. Compared to schemes relying solely on humidity or signal attenuation, this multi-source parameter fusion evaluation mechanism more accurately reflects the instantaneous changes in water mist along the radar propagation path and the operating status of the nozzle. This provides a reliable basis for subsequent dynamic threshold adjustments, ensuring that water mist filtration capacity and target detection sensitivity are precisely adapted to changes in cleaning conditions.

[0070] Based on any of the above embodiments, in Embodiment 4 of this application, referring to Figure 2 Before comparing the current distance value with the preset terrain simulation distance, the process includes: A10, obtains the current real-time attitude angles of the cleaning drone from the inertial measurement unit of the flight control system, the real-time attitude angles including pitch angle and roll angle.

[0071] In this embodiment, the inertial measurement unit refers to the combined sensor module of a three-axis gyroscope and a three-axis accelerometer integrated into the flight control system. The pitch angle refers to the angle of rotation about the horizontal axis in the UAV's body coordinate system, and the roll angle refers to the angle of rotation about the vertical axis in the UAV's body coordinate system. The real-time attitude angle refers to the filtered and fused attitude angle estimate output by the attitude calculation algorithm in each control cycle.

[0072] As an optional implementation, the main control chip of the flight control system reads the raw angular rate and acceleration data of the inertial measurement unit at a frequency of 1 kHz through a serial peripheral interface. The internal attitude calculation task runs a complementary filtering algorithm to perform weighted fusion of the attitude angles obtained by gyroscope integration and those obtained by accelerometer calculation, outputting updated pitch and roll angles. This attitude calculation task is executed once every 1 millisecond, and the generated attitude angles are stored in the global attitude structure in the flight control system's memory. The radar data processing process directly reads the latest pitch and roll angle values ​​from this structure in each control cycle for subsequent calculation of the composite tilt angle.

[0073] As an alternative implementation, the flight control system employs an extended Kalman filter as the attitude estimation algorithm, using quaternions as the state vector. It fuses angular rate, acceleration, and optional magnetometer data from the inertial measurement unit to output quaternion attitude estimation results. When acquiring real-time attitude angles, the flight control system converts the current quaternion into Euler angles, extracting the pitch and roll components. Using quaternion attitude estimation avoids the gimbal lock-up problem during large-angle maneuvers with Euler angles, ensuring continuous and reliable attitude angle acquisition even when the UAV tilts significantly in strong winds.

[0074] A20 calculates the combined tilt angle of the cleaning drone based on the pitch angle, the roll angle, and a first preset formula.

[0075] In this embodiment, the composite tilt angle refers to a spatial angle composed of the pitch angle and roll angle, representing the angle between the vertical axis of the UAV body and the local gravity direction. This angle is used to quantify the degree of tilt of the UAV in any direction. The first preset formula refers to a mathematical expression pre-written into the flight control program that takes the pitch angle and roll angle as inputs and outputs the composite tilt angle.

[0076] As an optional implementation, the first preset formula adopts an inverse cosine form, specifically expressed as Theta_tilt = arccos(cos Theta_pitch * cos Theta_roll), where Theta_tilt is the composite roll angle, Theta_pitch is the pitch angle, and Theta_roll is the roll angle. After obtaining the pitch and roll angles from the attitude structure, the flight control system calls the floating-point arithmetic unit to calculate the product of their cosine values, and then obtains the composite roll angle through a lookup table or hardware-accelerated inverse cosine function. This formula, based on the orthogonality of the direction cosine matrix, directly gives the angle between the aircraft's vertical axis and the direction of gravity, with a clear physical meaning.

[0077] As an alternative implementation, the first preset formula adopts an approximate synthesis form, calculating the synthesized tilt angle by taking the square root of the sum of squares under the assumption of small angles, expressed as Theta_tilt = sqrt(Theta_pitch^2 + Theta_roll^2). When both the pitch and roll angles are less than preset angle thresholds, the flight control system directly uses this approximate formula to calculate the synthesized tilt angle, reducing the processor overhead of inverse cosine calculations. When either angle exceeds the threshold, the flight control system switches to the inverse cosine formula to ensure calculation accuracy. This hybrid strategy achieves a balance between computational efficiency and full-angle accuracy.

[0078] A30, based on the synthesized tilt angle and the second preset formula, the current distance value is used as the slant distance to calculate the updated current distance value.

[0079] In this embodiment, slant range refers to the distance between the UAV and the target object, measured along the beamline direction by the ground ranging radar output. The updated current distance value refers to the true vertical distance between the UAV and the target object obtained after geometric correction of the slant range. This vertical distance will replace the original slant range for subsequent terrain-following flight control decisions.

[0080] As an optional implementation, the second preset formula adopts a cosine solution form, expressed as H_true = R_slant * cos Theta_tilt, where H_true is the updated current distance value, i.e., the true vertical distance, R_slant is the current distance value output by the ground ranging radar, i.e., the slant range, and Theta_tilt is the composite tilt angle calculated in step A20. The flight control system substitutes the composite tilt angle output in step A20 and the slant range output by the radar into this formula, obtains the vertical distance through floating-point multiplication and cosine operation, and writes this vertical distance into the input variable of the terrain-following control module, overwriting the original slant range value. Subsequent altitude comparison and PID control for terrain-following flight are both performed based on this updated vertical distance.

[0081] As an alternative implementation, the second preset formula introduces an installation angle correction term based on the cosine calculation. When the ground ranging radar is not installed strictly perpendicular to the bottom surface of the UAV but has a preset installation angle, the second preset formula is corrected to H_true = R_slant * cos(Theta_tilt + Theta_offset), where Theta_offset is the installation angle between the radar beam axis and the normal to the bottom surface of the UAV. This angle is obtained through factory calibration and stored in the flight control parameter table. The flight control system first reads this installation angle, adds it to the composite tilt angle, and then calculates the cosine value to obtain the vertical distance after installation angle compensation. This method can still ensure altitude correction accuracy in scenarios where the UAV cannot achieve a strictly perpendicular radar installation due to structural limitations.

[0082] Optionally, the first preset formula is: θ tilt =arccos(cosθ pitch *cosθ roll ); Where, θ tilt To synthesize the tilt angle, θ pitch Let θ be the pitch angle. roll This refers to the roll angle; The second preset formula is: H true =R slant *cos*θ tilt ; Among them, H true R is the updated current distance value. slant This refers to the current distance value.

[0083] For example, in strong winds, the cleaning drone tilts to the right to counteract crosswinds. The flight control system obtains the current pitch angle as -5 degrees and roll angle as 20 degrees from the inertial measurement unit. Using the inverse cosine formula, the combined tilt angle is calculated: cos(-5 degrees) is approximately 0.9962, cos(20 degrees) is approximately 0.9397, and their product is approximately 0.9361. The inverse cosine yields a combined tilt angle of approximately 20.6 degrees. At this time, the slant range output by the ground ranging radar is 1.65 meters. According to the cosine calculation formula, the vertical distance equals 1.65 multiplied by cos(20.6 degrees), i.e., 1.65 multiplied by 0.9361, which is approximately 1.54 meters. This vertical distance is sent as the updated current distance value to the terrain-following control module. Compared with the set distance of 1.5 meters, the difference is only 0.04 meters, and the flight control system does not execute an ascent maneuver. Without attitude compensation, directly comparing with a slant distance of 1.65 meters will result in a deviation of 0.15 meters, causing the drone to ascend incorrectly, wasting energy and affecting the uniformity of cleaning.

[0084] This embodiment acquires real-time attitude angles from the inertial measurement unit and synthesizes them into tilt angles. The tilt angles are then used to geometrically correct the radar slant range, eliminating terrain-following altitude measurement errors caused by the UAV's tilted flight under strong wind and crosswind conditions. This ensures the flight control system always controls altitude based on the actual vertical distance. This attitude compensation mechanism requires no additional hardware; it relies solely on existing attitude data and mathematical calculations from the flight control system. It can maintain the accuracy and safety of terrain-following flight even when the UAV is tilted significantly, effectively avoiding malfunctions caused by ranging errors.

[0085] Based on any of the above embodiments, in Embodiment 5 of this application, after calculating the combined tilt angle of the cleaning drone based on the pitch angle, the roll angle, and the first preset formula, the process includes: B10, when the absolute value of the synthesized tilt angle exceeds the preset tilt threshold, the SNR threshold and / or the RCS threshold are increased to reduce the interference of water mist density distribution changes on ranging accuracy under crosswind conditions.

[0086] In this embodiment, the preset tilt threshold refers to the critical composite tilt angle value used to determine whether the tilt degree of the UAV is sufficient to cause a significant change in the water mist distribution. Crosswind condition refers to the flight state of the UAV tilting its body towards the windward direction to maintain its position under the action of crosswind. At this time, the water mist is blown to one side by the crosswind, forming an asymmetrical distribution on the radar beam path.

[0087] As an optional implementation, the flight control system compares the absolute value of the synthesized tilt angle with a preset tilt threshold in each control cycle. This tilt threshold, calibrated through flight testing, is set at the critical tilt angle at which the water mist distribution begins to exhibit significant directionality. When the absolute value of the synthesized tilt angle is less than or equal to the tilt threshold, the water mist is approximately uniformly distributed along the radar beam path, and the current SNR and RCS thresholds remain unchanged, determined by the water mist concentration. When the absolute value of the synthesized tilt angle exceeds the tilt threshold, it indicates that the UAV is tilting significantly due to crosswinds, causing the water mist to be blown to one side of the aircraft. This results in a local increase in the density of the water mist layer through which the radar beam passes. In this case, the flight control system increases the SNR threshold by a first preset increment and the RCS threshold by a second preset increment. The updated SNR and RCS thresholds are written to the threshold register of the radar signal processing module to compensate for the increased target recognition difficulty caused by the local increase in water mist density.

[0088] As an alternative implementation, after detecting that the absolute value of the composite tilt angle exceeds a preset tilt threshold, the flight control system does not use a fixed incremental adjustment. Instead, it takes the difference between the absolute value of the composite tilt angle and the tilt threshold as input to a preset threshold gain mapping function. This function outputs an SNR threshold gain coefficient and an RCS threshold gain coefficient. The adjusted SNR threshold is obtained by multiplying the current SNR threshold by the SNR threshold gain coefficient, and the adjusted RCS threshold is obtained by multiplying the current RCS threshold by the RCS threshold gain coefficient. The gain mapping function is designed so that the larger the difference, the larger the gain coefficient, making the threshold increase positively correlated with the UAV's tilt degree, achieving dynamic threshold adjustment that adapts to crosswind intensity. After the threshold is updated, the flight control system continuously monitors the composite tilt angle. When the absolute value of the composite tilt angle falls below the tilt threshold and remains below it for a preset number of control cycles, the SNR threshold and RCS threshold are restored to their original values ​​before adjustment.

[0089] For example, when a cleaning drone operates in a level 6 crosswind, the aircraft tilts to the right to counteract the wind's effect. The flight control system calculates the current composite tilt angle to be 22 degrees. The preset tilt threshold is 15 degrees, the current SNR threshold is 15 dB, and the RCS threshold is 0.04 square meters. Since 22 degrees exceeds 15 degrees, the control system increases the SNR threshold from 15 dB to 18 dB and the RCS threshold from 0.04 square meters to 0.06 square meters. After the thresholds are increased, the radar effectively suppresses the dense water mist clutter that accumulates on the right beam path due to the crosswind, while the true echo from the photovoltaic panel can still be stably detected. When the wind weakens and the composite tilt angle drops to 10 degrees for 2 seconds, the control system restores the SNR and RCS thresholds to their original values ​​of 15 dB and 0.04 square meters.

[0090] Optionally, at this time, the execution logic is as follows: at time n, acquire the echo signal of the ground ranging radar; adjust the signal-to-noise ratio (SNR) threshold and radar cross section (RCS) threshold according to the water mist concentration of the current cleaning operation, and extract target points in the echo signal that simultaneously satisfy the SNR threshold and the RCS threshold, outputting the current distance value of the target points to filter out water mist clutter interference. The current distance value represents the distance between the cleaning drone and the target object below its horizontal plane; then acquire the current real-time attitude angle of the cleaning drone from the inertial measurement unit of the flight control system, the real-time attitude angle including pitch angle and roll angle; calculate the composite tilt angle of the cleaning drone based on the pitch angle, the roll angle, and a first preset formula. When the absolute value of the composite tilt angle exceeds a preset tilt threshold, increase the SNR threshold and / or the RCS threshold to reduce the interference of water mist density distribution changes on ranging accuracy under crosswind conditions. Then, based on the adjusted SNR threshold and / or RCS threshold, target points in the echo signal that simultaneously satisfy both the SNR threshold and the RCS threshold are extracted, and the current distance value of the target points is output. Then, based on the synthesized tilt angle and the second preset formula, the current distance value is used as the slant range to calculate an updated current distance value. Next, the current distance value is compared with a preset terrain-following distance. When the current distance value is less than the terrain-following distance, the cleaning drone is controlled to ascend to approach the terrain-following distance.

[0091] This embodiment directly correlates the synthetic tilt angle with the adjustment of the SNR and RCS thresholds, enabling the radar signal processing parameters to respond not only to the water mist concentration itself but also to changes in the spatial distribution of water mist caused by crosswind tilt. This mechanism solves the problem that water mist filtration schemes are effective in windless conditions but degrade in crosswind conditions due to the non-uniform distribution of water mist, ensuring that terrain-following ranging maintains reliable accuracy under complex wind field conditions.

[0092] Based on any of the above embodiments, in Embodiment Six of this application, after calculating the combined tilt angle of the cleaning drone based on the pitch angle, the roll angle, and the first preset formula, the process includes: C10, when the absolute value of the synthesized tilt angle exceeds the preset limit safety threshold, it is determined that the current flight attitude has exceeded the safety compensation range of terrain-following flight.

[0093] In this embodiment, the preset limit safety threshold refers to the critical composite tilt angle value that is pre-calibrated and characterizes the effective upper limit of the attitude compensation mechanism. The safety compensation range refers to the tilt angle range that can still ensure the accuracy and safety of terrain-following flight after geometric correction of the radar slant range by attitude angle. If the range is exceeded, the cosine calculation error will be amplified and the power margin of the UAV will be significantly insufficient, and there will be a collision risk if the terrain-following flight continues.

[0094] As an optional implementation, the flight control system compares the absolute value of the synthesized tilt angle with a preset safety threshold in each control cycle. This safety threshold is determined through flight envelope testing and is set as the tilt angle at which the attitude compensation error of the UAV begins to exceed the safety margin under the given radar beam angle and installation geometry conditions. When the absolute value of the synthesized tilt angle is less than or equal to the safety threshold, the current flight attitude is determined to be within the safety compensation range, and the terrain-following flight control logic continues to operate normally. When the absolute value of the synthesized tilt angle is greater than the safety threshold, the flight control system determines that the current flight attitude has exceeded the safety compensation range and immediately sets the terrain-following safety over-limit flag in the global status register. Setting this flag triggers the emergency response procedure described in step C20.

[0095] As an alternative implementation, the flight control system introduces a time-duration condition when determining whether the absolute value of the composite tilt angle exceeds the safety threshold. When the absolute value of the composite tilt angle exceeds the safety threshold, an over-limit timer is first started. If the absolute value of the composite tilt angle continues to exceed the safety threshold for a preset number of control cycles, it is determined that the flight attitude stability has exceeded the safety compensation range, and the terrain-following safety over-limit flag is set. If the absolute value of the composite tilt angle falls back below the safety threshold during the over-limit timer's counting period, the timer is reset, and the flag is not set. This mechanism is used to exclude situations where the attitude briefly exceeds the limit due to a sudden gust of wind and then recovers on its own, avoiding unnecessary emergency responses triggered by occasional transient disturbances.

[0096] C20 controls the cleaning drone to perform hovering or emergency landing and to terminate the current cleaning operation.

[0097] In this embodiment, hovering refers to the UAV maintaining its three-dimensional spatial position and stable attitude while in flight. Emergency landing refers to the UAV abandoning its current operational route and descending vertically to the ground or the nearest safe landing site along a preset safety strategy. Aborting the current cleaning operation means stopping the nozzles from spraying water, turning off the water pumps, exiting the terrain-following flight mode, and abandoning the unfinished cleaning route task.

[0098] As an optional implementation, the flight control system immediately executes an emergency response sequence. First, it shuts off the nozzle solenoid valve and stops the water pump motor output, terminating the spraying of cleaning fluid. Second, it forcibly switches the flight mode from terrain-following mode to absolute altitude hold mode, locking the current altitude using the current barometric altimeter or satellite positioning altitude as a reference, thus achieving hovering. While hovering, the flight control system continuously monitors the composite tilt angle. If the absolute value of the composite tilt angle falls below the safety threshold and remains stable within a preset waiting time, it is determined that the strong winds and gusts have passed, and the system determines whether to resume cleaning operations using remote control commands or preset logic. If the absolute value of the composite tilt angle fails to fall below the safety threshold within the preset waiting time, the system initiates an emergency landing procedure. During the emergency landing procedure, the flight control system calculates the return path based on the current flight path position and the coordinates of the preset safe landing point. It then controls the UAV to descend vertically to a preset altitude above the safe landing point at a constant descent rate before performing a touchdown landing. After landing, the flight control system locks all motors and records the operation interruption log.

[0099] As an alternative implementation, the flight control system executes an emergency landing directly without hovering. This strategy is suitable for scenarios where the extreme safety threshold setting has fully considered the critical conditions of the power margin. Once this threshold is exceeded, it indicates that the UAV no longer has sufficient power to maintain safe flight, and any pause increases the risk of loss of control. In this case, after shutting down the nozzles and water pumps, the flight control system directly calls the emergency landing subroutine to vertically land the UAV to the ground at the highest safe descent rate. Simultaneously, the flight control system writes the current operation interruption point's route number, geographical coordinates, remaining water volume, and fault cause code to the onboard non-volatile storage medium and triggers the onboard audible and visual alarm module to issue an operation interruption warning signal to ground personnel. After landing is complete and the motors are locked, the flight control system sends a telemetry data packet containing the interruption cause and location to the ground station via a data radio, allowing maintenance personnel to assess whether the operation can be resumed or whether manual intervention is required.

[0100] For example, a cleaning drone was operating in a sustained crosswind of level 7. The absolute value of its composite tilt angle reached 38 degrees, exceeding the preset safety threshold of 35 degrees, and remained above 38 degrees for six consecutive control cycles. The flight control system determined that the current attitude exceeded the safety compensation range, immediately shutting off the nozzles and water pumps, switching the flight mode from terrain-following to absolute altitude hold, and hovering the drone at the current altitude. After hovering for 5 seconds, the composite tilt angle increased to 42 degrees due to further wind force. The flight control system determined that the strong wind showed no signs of abating and immediately initiated an emergency landing procedure, descending vertically to a safe landing point at a rate of 2 meters per second, locking the motors, and sending an operation interruption alarm message and the coordinates of the interruption point to the ground station. Ground maintenance personnel dispatched manual inspection based on this information, and after the wind subsided, remotely controlled the drone to resume operation from the interruption point.

[0101] This embodiment constructs the final safety line of defense for the terrain-following flight control logic by forcibly triggering hovering or emergency landing and terminating operations when the synthesized tilt angle exceeds the ultimate safety threshold. This mechanism introduces a time-duration condition into the judgment criteria to distinguish between instantaneous disturbances and sustained exceedances. In terms of handling strategies, it provides two graded response paths: hovering for recovery and direct emergency landing. This ensures that even when attitude compensation fails due to extreme wind conditions, collisions will not occur due to the control logic continuing to execute terrain-following flight, fundamentally guaranteeing the flight safety and equipment integrity of the cleaning drone under high-wind conditions.

[0102] Based on any of the above embodiments, in Embodiment 7 of this application, after comparing the current distance value with the preset terrain-following setting distance, the process includes: D10, obtain the pre-stored flight path information, which indicates the location of the gaps between the photovoltaic arrays.

[0103] In this embodiment, the pre-stored flight path information refers to the waypoint sequence file generated by the flight path planning software and written into the flight control system memory before the cleaning operation. Each waypoint contains information such as latitude and longitude coordinates, altitude value, and waypoint attribute flag bits. The position of the photovoltaic array gap refers to the geographical coordinates of a specific waypoint marked as the gap entry point or gap exit point in the waypoint attribute flag bits, which characterizes the start and end boundaries of the channel gap between rows of photovoltaic modules.

[0104] As an optional implementation, during the route planning phase, maintenance personnel use ground station software to draw cleaning routes along the photovoltaic array layout direction on high-precision satellite imagery or a digital surface model of the power station, and manually mark the start and end points of the gap areas at the gap channels between photovoltaic arrays. The ground station software compiles the waypoint sequence and gap marking information into a binary route file. In this file, each waypoint data structure includes latitude, longitude, and altitude fields, and also allocates an additional one-byte waypoint attribute field, where the gap entry point is marked as 0x01, the gap exit point as 0x02, and the ordinary cleaning point as 0x00. Before the operation, this route file is loaded into the flight control system memory via a data transmission link or storage medium. During the terrain-following flight, the flight control system reads and caches the attribute information of the subsequent preset number of waypoints within the current flight segment in the order of waypoints.

[0105] As an alternative implementation, the gap locations in the flight path information are not manually marked, but are automatically identified and recorded through pre-executed reconnaissance flights. During the reconnaissance flight, the UAV, equipped with visual sensors or lidar, flies over the photovoltaic array at a safe altitude. The onboard edge computing unit identifies the start and end boundaries of each row of photovoltaic panels using a photovoltaic panel edge detection algorithm, and writes the detected gap center point coordinates, gap width, and gap extension direction into a gap feature database. The flight path planning algorithm uses this database as input to automatically generate a flight path file containing gap attribute annotations. Compared with manual annotation, this automatic annotation method can quickly update the gap information by re-executing reconnaissance flights after adjustments to the power plant component layout.

[0106] D20: When the cleaning drone is determined to be above the gap based on the flight path information, the nozzle is controlled to close.

[0107] In this embodiment, "about to enter the gap" refers to a state where the distance along the flight path between the UAV's current position and the next gap entry point in the flight path information is less than a preset early closure distance threshold. Early closure means cutting off the water supply from the nozzles before the UAV actually reaches the gap boundary, so that the water mist has basically dissipated before entering the gap, avoiding water waste and the risk of splashing onto electrical equipment that may be present in the gap.

[0108] As an optional implementation, the flight control system reads the current latitude and longitude of the UAV output by the Global Navigation Satellite System receiver in each control cycle and calculates the spherical distance one by one with the coordinates of subsequent waypoints cached in the flight path information. When it detects that the estimated remaining distance corresponding to the difference between the current waypoint number and the waypoint number of the next gap entry point multiplied by the waypoint spacing is less than the preset early closing distance, the flight control system sends a low-level signal to the nozzle solenoid valve drive circuit through the general input / output interface. The solenoid valve coil is de-energized, the valve core closes the flow channel under the action of the return spring, and the nozzle stops spraying water. The early closing distance is determined by adding a safety margin to the product of the UAV's current flight speed and the solenoid valve response time, ensuring that the nozzle is completely closed when the UAV reaches the gap boundary.

[0109] As an alternative implementation, the flight control system does not rely on waypoint numbers to calculate the remaining distance, but instead uses a geofencing triggering method. A virtual vertical plane at a preset pre-closing distance ahead of each gap entry point in the flight path information is defined as the closing trigger fence. The flight control system continuously detects the UAV's real-time latitude and longitude against all closing trigger fences within polygons. When the UAV crosses any closing trigger fence, it triggers the nozzle shutdown interruption service routine, which immediately sets the nozzle shutdown flag and executes a solenoid valve de-energization operation. This method utilizes the precise boundary triggering mechanism of geofencing to avoid lead deviations in waypoint calculation caused by fluctuations in flight speed or changes in flight path curvature.

[0110] D30: When it is determined from the flight path information that the cleaning drone has crossed the gap and is once again above the photovoltaic panel, the nozzle is restarted to resume the current cleaning operation.

[0111] In this embodiment, "having crossed the gap and being above the photovoltaic panel again" means that the UAV's current position has passed the gap departure point marked in the flight path information and entered the area directly above the next row of photovoltaic panels along the flight path direction. "Reopening the nozzle" means that the solenoid valve coil is re-energized, the valve core opens, and high-pressure water flows through the nozzle and is atomized and sprayed onto the surface of the photovoltaic panel.

[0112] As an optional implementation, after detecting that the UAV's current waypoint number has exceeded the gap departure point waypoint number, the flight control system continues to monitor the distance along the flight path between the UAV's real-time position and the gap departure point. When this distance is greater than a preset delayed opening distance, it is determined that the UAV has safely entered the area above the photovoltaic panel. The flight control system sends a high-level signal to the nozzle solenoid valve drive circuit, the solenoid valve opens, the water pump output returns to normal cleaning power, and the current cleaning operation continues. The delayed opening distance setting ensures that the nozzle resumes spraying water only after the UAV has completely left the gap area and the UAV is stable above the photovoltaic panel, avoiding some water mist drifting into the gap due to positioning errors at the gap boundary.

[0113] As an alternative implementation, a radar echo signal-assisted confirmation mechanism is introduced into the nozzle restart decision. When the flight control system determines, based on the flight path information, that the UAV waypoint has crossed the gap departure point, it does not immediately activate the nozzle. Instead, it waits for the current distance value output by the ground ranging radar to recover to the preset tolerance range of the terrain-following distance and remain there for a preset number of control cycles before executing the nozzle activation action. This dual confirmation mechanism utilizes the real-time capability of radar ranging to cross-verify the gap markings in the flight path information. Only when both the flight path information and the radar's measured distance indicate that the UAV is above the photovoltaic panel does the flight control system output an activation signal to the solenoid valve. This ensures that the nozzle will not be mistakenly activated above the gap or at the edge of the photovoltaic panel if there is a deviation in the flight path information or if the UAV's actual trajectory deviates from the preset flight path.

[0114] For example, a cleaning drone flies from north to south along a preset route to clean the third row of photovoltaic panels. The flight control system caches waypoint 58 as the gap entry point and waypoint 62 as the gap exit point in the next flight segment, with the distance between the two points corresponding to a 4-meter gap channel on the ground. When the drone flies past waypoint 55, the flight control system calculates that the remaining distance is approximately 3 waypoints, which is less than the preset pre-closing distance of 2 meters. Therefore, it controls the nozzle solenoid valve to de-energize, and the nozzle closes. The nozzle remains closed while the drone flies over the gap at a ground-mimicking altitude of 1.5 meters. After flying past waypoint 62, the flight control system detects that the waypoint number has passed the gap exit point, and the current distance value output by the ground ranging radar is stable within the range of 1.5 meters ± 0.1 meters for 5 consecutive control cycles. It determines that the drone has safely entered the area above the fourth row of photovoltaic panels, and then outputs an open signal to the solenoid valve, the nozzle resumes spraying water, and the cleaning operation continues.

[0115] This embodiment achieves time synchronization between the cleaning operation and the array terrain features by pre-marking the location of the photovoltaic array gaps in the flight path information and closing the nozzles before flying over the gaps and activating them with a delay after flying over the gaps. This mechanism further solves the problem of ineffective water spraying above the gaps, reducing cleaning fluid waste and avoiding potential impacts of water mist on electrical equipment at the gaps, based on the safety strategy of ascending without descending during terrain-following flight. The introduction of a radar echo cross-verification activation confirmation mechanism further improves the reliability of nozzle control at the gap boundaries and reduces the risk of accidental spraying due to flight path deviations.

[0116] Based on any of the above embodiments, in Embodiment 8 of this application, before comparing the current distance value with the preset terrain-following distance, the process includes: The system acquires wind direction and speed information of the current operating environment, as well as the real-time heading angle of the cleaning drone. It calculates the crosswind component acting on the cleaning drone based on the wind direction, wind speed, and real-time heading angle, and determines the water mist offset direction and water mist offset intensity based on the crosswind component. It then verifies the reliability of the current distance value based on the water mist offset direction and water mist offset intensity. If the current distance value is deemed reliable, it is sent to subsequent comparison steps. If the current distance value is deemed unreliable, it is compensated and corrected before being sent to subsequent comparison steps.

[0117] In this embodiment, the crosswind component refers to the projected amplitude of the wind speed vector in the direction perpendicular to the UAV's heading. This component characterizes the intensity of the lateral wind's effect on the water mist. The water mist offset direction refers to the main direction in which the water mist is blown under the action of the crosswind, which is divided into left offset and right offset relative to the UAV's body coordinate system. The water mist offset intensity is a quantitative indicator of the degree of water mist accumulation in the offset direction, which is positively correlated with the crosswind component and negatively correlated with the nozzle atomization diffusion angle. The credibility verification refers to the process of assessing whether the radar output current distance value under the current water mist offset conditions is the true surface echo of the photovoltaic panel.

[0118] As an optional implementation, the flight control system acquires the average wind speed and average wind direction of the current operating environment through airborne meteorological wind sensors or meteorological data transmitted from ground stations. Simultaneously, it reads the UAV's real-time heading angle from the attitude reference system fused with the inertial measurement unit and magnetometer. First, the absolute value of the difference between the wind direction and the heading angle is calculated to obtain the wind angle difference. When the wind angle difference is within the range of 60 to 120 degrees or 240 to 300 degrees, a significant crosswind is determined. The wind speed value is multiplied by the sine of the wind angle difference to obtain the crosswind component value. Then, the crosswind component value and a preset nozzle atomization diffusion angle parameter are substituted into a water mist offset intensity mapping function. This mapping function outputs a water mist offset intensity level; a higher level indicates more severe water mist accumulation in the crosswind direction. Simultaneously, the water mist offset direction is determined based on the quadrant of the wind angle difference. After acquiring the water mist offset direction and intensity, the reliability of the current distance value output in step S20 is verified. The specific method for reliability verification is as follows: Read the echo amplitude values ​​of the ground ranging radar on the left and right sides of the M nearest range cells corresponding to the current range value. When the water mist offset direction is leftward, focus on checking whether the echo amplitude value of the left-side neighboring range cells shows an abnormal increase. If the proportion of the amplitude value in the left-side neighboring cells exceeding the nominal echo amplitude of the photovoltaic panel exceeds a preset proportion, it is determined that there is a dense water mist mass on the left, and the target point corresponding to the current range value may be the edge of the water mist mass rather than the actual surface of the photovoltaic panel, thus determining that the current range value is unreliable. When the current range value is determined to be unreliable, compensation correction is performed on the current range value. The compensation correction amount is the water mist offset intensity level multiplied by a preset unit correction step size, and the correction direction is adjusted to the right, i.e., away from the water mist offset direction, to approximate the actual surface position of the photovoltaic panel. When no abnormal amplitude increase occurs in the left-side neighboring range cells, the current range value is determined to be reliable, and the process is directly sent to step S30.

[0119] As an alternative implementation, the flight control system does not rely on independent wind sensors but indirectly infers the degree of crosswind influence using the UAV's own flight status. In each control cycle, the flight control system acquires the roll direction correction torque command value output by the attitude controller. This command value represents the roll torque required by the UAV to maintain its position against crosswinds, and its amplitude is positively correlated with the crosswind component. Dividing the roll direction correction torque command value by a preset maximum torque value yields a normalized crosswind intensity factor, which implicitly includes wind speed information and the UAV's windward area information. The normalized crosswind intensity factor is then input into a nonlinear mapping table for water mist bias intensity. This mapping table, calibrated experimentally, outputs the water mist bias intensity level under the current operating conditions. The water mist bias direction is determined by the sign of the roll torque command; a positive sign corresponds to left bias, and a negative sign corresponds to right bias. The reliability verification method involves reading the distance value sequence output by the ground ranging radar within the most recent K consecutive frequency modulation cycles, calculating the variance of the sequence, and the maximum jump variable of the difference between adjacent cycles. When the water mist bias intensity level exceeds a preset level and the variance of the range value sequence exceeds a preset variance threshold, it is determined that the water mist bias has caused the radar to alternately lock onto the water mist clusters and the photovoltaic panel surface across multiple frames, and the range value output in the current cycle is in an unreliable state. At this time, compensation and correction are performed on the current range value, using the median of the range value sequence as the correction value. The median has natural robustness to outliers caused by occasional water mist cluster locking, and can achieve software-level stability of the range value without changing the radar hardware and signal processing algorithm. When the variance of the range value sequence does not exceed the preset variance threshold, the current range value is determined to be reliable, and the process is directly sent to step S30.

[0120] For example, a cleaning drone flies along an east-west route in a Force 6 left-hand wind, with a heading angle of 90 degrees, an ambient wind direction of 0 degrees (north), and a wind speed of 12 meters per second. The wind angle difference is 90 degrees, indicating a significant crosswind, with a crosswind component value of 12 multiplied by sin(90 degrees), equaling 12 meters per second. The water mist is deflected to the left, meaning the water mist is blown towards the left side of the aircraft, and the water mist deflection intensity is mapped to a high level. The radar output shows a current distance of 1.38 meters, less than the set distance of 1.5 meters. Without a reliability check, an ascent command will be triggered. The flight control system reads the echo amplitude values ​​of the five distance cells adjacent to the current distance value on its left. It finds that the amplitude values ​​of three of these cells exceed 80% of the nominal echo amplitude of the photovoltaic panel, indicating the presence of a dense water mist mass on the left. The current distance value of 1.38 meters may be a false echo from the leading edge of the water mist mass and is therefore deemed unreliable. The flight control system corrected the current distance value by 0.15 meters to the right, thus deviating from the water mist offset direction, resulting in a corrected distance value of 1.53 meters. After being sent to the comparison step, it was determined that this distance was greater than the terrain-following set distance, and the current altitude was maintained without increasing. During subsequent flight, the crosswind weakened, the water mist offset intensity decreased to a low level, the variance of the distance value sequence returned to normal, the distance value verification passed, and the UAV resumed normal terrain-following control.

[0121] This embodiment addresses the issue of radar potentially locking onto water mist clumps and outputting erroneous distance values ​​when crosswinds cause asymmetrical water mist distribution by introducing a reliability verification and compensation correction step based on crosswind perception after distance value acquisition and before terrain simulation comparison. This mechanism extracts the impact of crosswinds on water mist distribution in real time from wind field information or flight status, determines the reliability of the current distance value based on the amplitude analysis of neighboring distance cells or the statistical characteristics of the distance value sequence, and applies directional or statistical compensation when the value is unreliable. This prevents erroneous distance values ​​from propagating to the flight control decision-making process, ensuring that only distance values ​​accurately reflecting the photovoltaic panel surface participate in terrain simulation comparison. This maintains the accuracy and safety of terrain simulation flight even under complex conditions of crosswind and water mist coupling interference.

[0122] This application provides a control device for a cleaning drone, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the control method for the cleaning drone in the first embodiment described above.

[0123] The following is for reference. Figure 3 The diagram illustrates a structural schematic of a control device suitable for implementing the embodiments of the present application for a cleaning drone. The control device for the cleaning drone in the embodiments of the present application may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, personal digital assistants (PDAs), tablets, and in-vehicle terminals, as well as fixed terminals such as digital TVs and desktop computers. Figure 3 The control device for the cleaning drone shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0124] like Figure 3As shown, the control device for the cleaning drone may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to programs stored in read-only memory (ROM) 1002 or programs loaded from storage device 1003 into random access memory (RAM) 1004. The random access memory 1004 also stores various programs and data required for the operation of the cleaning drone's control device. The processing unit 1001, ROM 1002, and RAM 1004 are interconnected via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to I / O interface 1006: input devices 1007 including, for example, touchscreens, touchpads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output devices 1008 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 1003 including, for example, magnetic tapes, hard disks, etc.; and communication devices 1009. Communication device 1009 allows the control equipment of the cleaning drone to communicate wirelessly or wiredly with other devices to exchange data. Although the figure shows control equipment for a cleaning drone with various systems, it should be understood that it is not required to implement or possess all the systems shown. More or fewer systems can be implemented alternatively.

[0125] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from ROM 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0126] The control device for cleaning drones provided in this application, employing the control method for cleaning drones in the above embodiments, can solve the technical problem of abnormal drone descent caused by gaps in photovoltaic panels. Compared with the prior art, the beneficial effects of the control device for cleaning drones provided in this application are the same as those of the control device for cleaning drones provided in the above embodiments, and other technical features in the control device for cleaning drones are the same as those disclosed in the method of the previous embodiment, and will not be repeated here.

[0127] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0128] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0129] This application provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, the computer-readable program instructions being used to execute the control method for the cleaning drone in the above embodiments.

[0130] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems or devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), flash memory, optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, radio frequency (RF), or any suitable combination thereof.

[0131] The aforementioned computer-readable storage medium may be included in the control equipment of the cleaning drone; or it may exist independently and not be assembled into the control equipment of the cleaning drone.

[0132] The aforementioned computer-readable storage medium carries one or more programs. When the one or more programs are executed by the control device of the cleaning drone, the control device of the cleaning drone causes the following: acquiring the echo signal of the ground ranging radar; adjusting the signal-to-noise ratio (SNR) threshold and radar cross section (RCS) threshold according to the water mist concentration of the current cleaning operation, and extracting target points in the echo signal that simultaneously meet the SNR threshold and RCS threshold conditions, outputting the current distance value of the target points to filter out water mist clutter interference, wherein the current distance value represents the distance between the cleaning drone and the target object below its horizontal plane; comparing the current distance value with a preset terrain-following distance, and when the current distance value is less than the terrain-following distance, controlling the cleaning drone to ascend to approach the terrain-following distance.

[0133] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0134] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation that may be implemented in systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may be executed in a different order than that shown in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0135] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0136] The readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the control method of the cleaning drone described above, which can solve the technical problem of abnormal drone sinking caused by gaps in photovoltaic panels. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as the beneficial effects of the control method of the cleaning drone provided in the above embodiments, and will not be repeated here.

[0137] This application provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the control method for cleaning drones as described above.

[0138] The computer program product provided in this application can solve the technical problem of abnormal descent of drones caused by gaps in photovoltaic panels. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as the beneficial effects of the control method for cleaning drones provided in the above embodiments, and will not be repeated here.

[0139] The above are merely preferred embodiments of this application and do not limit the patent scope of this application. Any equivalent structural or procedural transformations made using the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are included within the patent protection scope of this application.

Claims

1. A control method for a cleaning drone, characterized in that, This method is applied to a cleaning drone, which is equipped with a ground ranging radar, and includes the following control methods: Acquire the echo signal of the ground ranging radar; The system obtains the air humidity of the current working environment collected by the humidity sensor on the cleaning drone, and obtains the water pump output power of the cleaning drone in the current cleaning operation. The echo signal attenuation is calculated based on the reference echo signal strength of the ground ranging radar when the nozzle is not turned on and the echo signal strength after the nozzle is turned on. The water mist concentration is calculated based on the air humidity, the water pump output power, and the echo signal attenuation, using a water mist evaluation formula. The water mist evaluation formula includes: Cmist=α*Hcurr+β*Ppump+γ*ΔSCmist Where α is the humidity weighting coefficient, β is the power weighting coefficient, γ is the attenuation weighting coefficient, Hcurr is the air humidity, Ppump is the water pump output power, ΔSCmist is the echo signal attenuation, and Cmist is the water mist concentration. When the water mist concentration exceeds a preset concentration threshold, the SNR threshold and / or RCS threshold are increased to enhance the filtering capability of water mist clutter. When the water mist concentration is lower than or equal to the preset concentration threshold, the SNR threshold and / or RCS threshold are reduced to the default value to improve the detection sensitivity of photovoltaic panel targets; Extract the target points in the echo signal that simultaneously satisfy the SNR threshold and the RCS threshold, and output the current distance value of the target points to filter out water mist noise interference. The current distance value represents the distance between the cleaning drone and the target object below its horizontal plane. The real-time attitude angles of the cleaning drone are obtained from the inertial measurement unit of the flight control system. The real-time attitude angles include pitch angle and roll angle. The combined tilt angle of the cleaning drone is calculated based on the pitch angle, the roll angle, and the first preset formula; Based on the synthesized tilt angle and the second preset formula, the current distance value is used as the slant distance to calculate the updated current distance value; The updated current distance value is compared with the preset terrain-following distance. When the current distance value is less than the terrain-following distance, the cleaning drone is controlled to ascend to approach the terrain-following distance. When the current distance value is greater than the ground-following set distance, the cleaning drone is controlled to maintain the current altitude.

2. The control method for a cleaning drone as described in claim 1, characterized in that, The first preset formula is: θtilt=arccos(cosθpitch*cosθroll); Where θtilt is the composite tilt angle, θpitch is the pitch angle, and θroll is the roll angle; The second preset formula is: Htrue = Rslant * cos * θtilt; Where Htrue is the updated current distance value, and Rslant is the current distance value.

3. The control method for a cleaning drone as described in claim 1, characterized in that, After calculating the composite tilt angle of the cleaning drone based on the pitch angle, the roll angle, and the first preset formula, the process includes: When the absolute value of the synthesized tilt angle exceeds the preset tilt threshold, the SNR threshold and / or the RCS threshold are increased to reduce the interference of water mist density distribution changes on ranging accuracy under crosswind conditions.

4. The control method for a cleaning drone as described in claim 1, characterized in that, After calculating the composite tilt angle of the cleaning drone based on the pitch angle, the roll angle, and the first preset formula, the process includes: When the synthesized tilt angle exceeds the preset limit safety threshold, it is determined that the current flight attitude has exceeded the safety compensation range of terrain-following flight. Control the cleaning drone to perform hovering or emergency landing, and suspend the current cleaning operation.

5. The control method for a cleaning drone as described in claim 1, characterized in that, After comparing the updated current distance value with the preset terrain setting distance, the process includes: Obtain pre-stored flight path information, wherein the positions of the photovoltaic array gaps are marked in the flight path information; When the cleaning drone is determined to be above the gap based on the flight path information, the nozzle is controlled to close. When the cleaning drone is determined to have crossed the gap and is now above the photovoltaic panel again based on the flight path information, the control nozzle is restarted to resume the current cleaning operation.

6. A control device for a cleaning drone, characterized in that, The control device of the cleaning drone includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the control method for the cleaning drone as described in any one of claims 1 to 5.

7. A storage medium, characterized in that, The storage medium is a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the steps of the control method for the cleaning drone as described in any one of claims 1 to 5.

Images