Control method of intelligent outdoor operation spraying unmanned aerial vehicle, spraying unmanned aerial vehicle and medium

Abstract

The invention relates to a control method of an intelligent outdoor operation spraying unmanned aerial vehicle, the spraying unmanned aerial vehicle and a medium in the technical field of unmanned aerial vehicles. The control method comprises the steps that wall face features are obtained through calculation according to a wall face image I shot by the unmanned aerial vehicle, and spraying pressure compensation and unmanned aerial vehicle corrected speed are calculated according to the wall face features. And calculating the initial spraying pressure based on the reference spraying pressure. And the interference factors of the unmanned aerial vehicle are compensated. And performing weighted fusion on the spraying pressure compensation, the spraying initial pressure and the interference factor compensation to obtain final spraying pressure. And the final spraying flow is calculated according to the corrected speed of the unmanned aerial vehicle. By means of a newly-designed spraying logic architecture, two-dimensional regulation and control of quality pre-judgment and system stability are achieved, the adaptability to complex working conditions is improved, the coating uniformity and the spraying quality are guaranteed, and the system has the advantages of being compact in structure, convenient and fast to operate and high in anti-interference capacity and is suitable for spraying operation of various outdoor walls, storage tanks and the like.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) technology, and in particular to a control method for an intelligent outdoor spraying UAV, the spraying UAV, and the spraying medium. Background Technology

[0002] For large outdoor structures, including building facades, bridges, and large storage tanks, surface painting currently relies primarily on two traditional methods: scaffolding or aerial work platforms, with manual spraying using handheld spray guns. These methods generally suffer from low efficiency, high safety risks, and an over-reliance on individual worker experience for painting quality. While some technologies have incorporated drones into painting operations, these drone-based technologies still have significant limitations. For example, existing solutions often employ a simple integration of a drone platform and painting equipment, with relatively rudimentary control logic. Most technologies simply apply general control or neural network models, often lacking adaptability for the specific scenario of painting. Furthermore, existing methods often linearly adjust spray pressure based solely on the distance between the drone and the wall. However, during painting operations, various dynamic disturbances in the outdoor environment, such as sudden gusts of wind, changes in wall roughness, and variations in paint viscosity with temperature, can lead to unstable spray quality from drones, easily causing defects, material waste, and rework. Summary of the Invention

[0003] To address the technical problem of unstable coating quality in existing drones, this invention provides a control method for an intelligent outdoor spraying drone, a spraying drone, and a coating medium.

[0004] In a first aspect, the present invention proposes a control method for an intelligent outdoor spraying drone, comprising: controlling a wall surface image captured by the drone. I The wall surface characteristics obtained by calculation: roughness R mean Texture complexity T complex Reflectivity R ref The wall surface is classified into different work surface levels based on its characteristics, and the spraying pressure compensation Δ corresponding to each work surface level is obtained through preset rules. P wall .according to T complex and the drone's base flight speed V base The corrected speed of the drone was calculated. v .

[0005] By comparing the drone's coordinate data with the preset path, the drone's trajectory deviation Δ is obtained.R If Δ R If the value is greater than or equal to the threshold, then the trajectory is corrected; if Δ R If the initial spraying pressure is less than the threshold, then calculate the initial spraying pressure. P init : P init = P base × F ; In the formula, P base As the reference spraying pressure, F To integrate decision coefficients, compensation is made for interference factors affecting the UAV: ​​if gust interference is detected, compensation is made based on the UAV's attitude offset Δ. i Calculate gust compensation Δ P wind If a difference Δ in the mass of the drone is detected... m If the value exceeds the threshold, then the value is based on the drone's current mass. m real Initial weight m 0. Nozzle spray area S Calculate load compensation Δ P load If the tension in the drone's pipeline is detected F pipe If greater than or equal to the threshold, then based on F pipe Calculate pipeline tension compensation Δ P pipe .

[0006] according to P init Δ P wall Δ P wind Δ P load Δ P pipe The final spraying pressure was calculated. P final .according to v The final spraying flow rate was calculated. Q .

[0007] Secondly, the present invention also proposes an intelligent outdoor spraying drone, comprising: a body, a spraying assembly, and a controller. The spraying assembly is used for spraying walls. The controller uses the control method for the intelligent outdoor spraying drone described in the first aspect to control the spraying pressure and spraying flow rate of the spraying assembly.

[0008] Thirdly, the present invention also proposes a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the control method for the intelligent outdoor spraying drone in the first aspect.

[0009] The beneficial effects of this invention are as follows: This invention achieves dual-dimensional control of quality prediction and system stability through a newly designed spraying logic architecture, improving adaptability to complex working conditions, ensuring coating uniformity and spraying quality, and has the advantages of compact structure, convenient operation, and strong anti-interference ability. It is suitable for spraying operations on various outdoor walls, storage tanks, etc. Attached Figure Description

[0010] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0011] Figure 1 This is a structural diagram of an intelligent outdoor spraying drone; Figure 2 This is a flowchart of the control method for intelligent outdoor spraying drones.

[0012] In the diagram: 1-1 Body; 1-2 Controller; 1-3 Spraying assembly; 1-4 Radar; 1-5 Industrial camera; 1-6 Quick-release interface; 1-7 Spraying pipe; 1-8 Spray nozzle. Detailed Implementation

[0013] The technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0014] It should be noted that when a component is said to be "installed on" another component, it can be directly on the other component or it may be in a component that is centered on it. When a component is said to be "set on" another component, it can be directly set on the other component or it may also be in a component that is centered on it. When a component is said to be "fixed to" another component, it can be directly fixed to the other component or it may also be in a component that is centered on it.

[0015] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the specification of this invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "or / and" as used herein includes any and all combinations of one or more of the associated listed items.

[0016] Please refer to Figure 1This embodiment provides an intelligent outdoor painting drone, which includes a fuselage 1-1, a painting assembly 1-3, an environmental perception module, and a controller 1-2. The fuselage 1-1 can be a multi-rotor structure, such as a hexacopter or octocopter model. The fuselage 1-1 is equipped with corresponding arms, frames, landing gear, battery compartments, power modules, flight control modules, power management modules, wireless communication modules, and other components. The fuselage 1-1 can be a directly selected model containing the above components, which will not be described in detail here. The painting assembly 1-3 includes: a storage tank, a gear pump, a regulating valve, a painting pipe 1-7, and a nozzle 8. The above structures are integrated on a bracket and connected to the fuselage 1-1 through a quick-release interface 1-6, so that the painting assembly 1-3 can be completely disassembled and assembled from the fuselage 1-1, forming a quick-release structural design. The painting pipe 1-7 can be made of metal, and its length is set according to the size of the fuselage 1-1, requiring the length to exceed the projection area of ​​the rotor. The nozzle 8 can be a duckbill nozzle 8 to increase the spraying area. The spraying assembly 1-3 uses a gear pump and regulating valve to spray the paint in the storage tank onto the wall / work surface through the spray pipe 1-7 and nozzle. Furthermore, a level sensor and a viscosity sensor can be installed in the storage pipe to collect the remaining paint amount and viscosity, respectively. After knowing the remaining paint amount, the current mass of the drone after spraying can be calculated by combining the paint density and the drone's mass. The environmental perception module is installed on the body 1-1 and includes: a millimeter-wave radar 1-4 for collecting the distance between the drone and the work surface; an industrial camera 1-5 for collecting image data of the work surface; an attitude sensor for collecting the drone's pitch angle, roll angle, yaw angle, and rate of attitude change; and an RTK (Real-Time Dynamic Carrier Phase Differential) positioning unit for real-time output of the drone's position (i.e., three-dimensional coordinates x, y, z). The controller 1-2 is installed on the body 1-1, forming an edge computing unit. The hardware of the controller 1-2 can be implemented using a microcontroller, embedded industrial computer, or automotive-grade computing platform. Its internal control logic can autonomously realize the drone's route planning, spraying pressure, and spraying flow rate adjustment, forming a closed-loop control. Route planning can employ common drone planning methods. Spraying pressure and flow rate adjustment control is one of the innovations of this invention; this adjustment control logic is the control method for the intelligent outdoor spraying drone of this invention. The controller 1-2 uses the intelligent outdoor spraying drone control method to control the spraying pressure and flow rate of the spraying assembly 1-3. Specifically, as shown... Figure 2 As shown, the control method for intelligent outdoor spraying drones includes: First, multi-source data acquisition is performed. The environmental perception module mounted on the drone 1-1 collects multi-source data, including but not limited to: the distance between the drone and the work surface. D (Can be acquired at a frequency of 10Hz) Wall image I UAV attitude data A(Including pitch angle, roll angle, yaw angle and their rate of change), UAV coordinate data L Coating viscosity or Current quality of drones m real The aforementioned multi-source data, after collection and preprocessing, undergoes standardized processing and time synchronization via hardware timestamps to ensure a synchronization error of ≤1ms, preventing data timing misalignment. For example, during preprocessing... D Noise reduction can be achieved using a sliding window filtering method (window length of 3). A , L Kalman filtering can be used for smoothing to reduce measurement errors. I A weighted average method can be used for grayscale processing. To adapt to the subsequent multi-source data fusion requirements, all preprocessed data are linearly normalized to the [0,1] interval to ensure uniform parameter dimensions and unbiased weight calculation.

[0017] After obtaining multi-source data, according to I The wall surface characteristics obtained by calculation: roughness R mean Texture complexity T complex Reflectivity R ref .in, R mean Sobel horizontal gradient method can be used to extract... I Mean grayscale gradient of a 5×5 window Then, through linear mapping, we obtain: .

[0018] .

[0019] In the formula, G x ( x , y )for I The middle position is ( x , y The Sobel horizontal gradient value of the pixel. k G The characteristic value is 0.002 in this embodiment. b G As a fixed coefficient, it is obtained by linear fitting of 50 sets of measured data of typical outdoor concrete, steel structure and other sprayed wall surfaces, and is adapted to the outdoor spraying operation scenario of this invention.

[0020] T complex Canny edge detection can be used for extraction.I After calculating the number of pixels at the edge of the 5x5 window, we get: .

[0021] In the formula, N edge for I The number of edge pixels. R ref This can be obtained by statistical analysis of grayscale values: .

[0022] In the formula, I ( x , y ( ) represents the pixel grayscale value. After obtaining the values ​​of these three specific wall features, the work surface level of the wall is classified based on the experimentally calibrated threshold. In this embodiment, the work surface levels are divided into: Level 3 (optimal), Level 2 (medium), and Level 1 (complex). Specifically, the classification method includes: like R mean ≤0.8∩ T complex ≤0.6∩ R ref If the value is ≥0.4, it is classified into three levels. 0.8 < R mean ≤1.5∩0.6< T complex ≤0.8∩0.3≤ R ref If <0.4, it is classified as level two. If R mean >1.5∩ T complex >0.8∩ R ref If the value is less than 0.3, it is classified as Level 1.

[0023] Subsequently, the spraying pressure compensation Δ corresponding to the work surface level is obtained through preset rules. P wall In this embodiment, the third-level working surface is Δ P wall =0.1MPa. For the secondary working face, Δ P wall =0.2MPa. For the first-level working face, Δ P wall =0.3MPa. Then according to... T complex and the drone's base flight speed V base The corrected speed of the drone was calculated. v : v = V base ×(1- k v × T complex ).

[0024] In the formula, k v The correction factor, derived from industry experience, is used to balance texture complexity and spraying efficiency.

[0025] On the other hand, after calculating Δ P wall , v At the same time, the coordinate data of the drone L Sequence and Preset Path L 预设 In comparison, the trajectory deviation Δ of the UAV is calculated based on the degree of deviation. R : Δ R =| L - L 预设 |

[0026] If Δ R If the value is greater than or equal to the threshold (5cm in this embodiment), the trajectory is corrected so that the drone path returns to the correct value. L 预设 If Δ R If the spraying pressure is less than 5cm, calculate the initial spraying pressure. P init : P init = P base × F .

[0027] In the formula, P base As the reference spraying pressure, F To integrate decision coefficients. It is worth mentioning that... F It is also based on real-time updates of collected multi-source data. Specifically, F The calculation method is as follows: F = α · D norm + β · I feature +g · A stab + d · L acc .

[0028] .

[0029] In the formula, D norm The optimal operating distance for drones. I feature for I The comprehensive image feature value is obtained by taking the arithmetic average of the three preprocessed wall features. A stab Let be the rate of change of attitude of the drone. L acc This represents the positioning error. α For distance data weights, β Image feature weights, c As the weights for the attitude data, d To determine the weights of the data. Among them, D norm For preprocessed D With optimal working distance D opt After linear normalization to the baseline, it can be expressed as: .

[0030] same, A stab It can be obtained from pre-processed A The pitch / roll rate of change is extracted, linearly normalized, and calculated using the following formula: .

[0031] The maximum allowable rate of change in the formula is preset to 0.5° / s, which is a common empirical value for outdoor drone spraying processes. L acc For preprocessed L The 3D positioning error relative to the preset working coordinates is normalized based on the maximum permissible error of 0.01m for RTK positioning, and the calculation formula is as follows: .

[0032] for α , β , c , d A genetic algorithm can be used for optimization, with the objective being to optimize the weighted sum of the comprehensive errors. J Minimum, with the following constraints: α + β + c + d =1; where the optimization objective function is: min J=0.4Δ Q+ 0.3Δ D+ 0.3Δ A .

[0033] In the formula, different weight combinations are tried using a genetic algorithm, and the results are calculated by substituting the values ​​into the formula. J value, J The smaller the value, the better the weight combination, and Δ Q / Δ D / Δ A It is a calculation J The three core deviation terms of the value vary with the weight combination. These three parameters are all non-negative deviation values ​​(the larger the deviation, the worse the operational status in that dimension), corresponding to the three dimensions of spraying operation: quality, distance, and attitude. They are directly related to the positioning data and four weights (the higher a certain type of weight, the greater the impact of the corresponding deviation term on the overall error). Among them, Δ... Q To follow β The relevant coating quality deviation (unit: μm) is the absolute deviation between the actual coating thickness predicted based on wall characteristics and the target thickness of the spraying process (the larger the deviation, the more likely the coating is to accumulate or have missed areas, resulting in poorer quality). Δ Q The calculation formula is: Δ Q =|Predicted Coating Thickness - Target Process Thickness|. The target process thickness can be set to 80μm (an industry-standard optimal value). The predicted coating thickness can be calculated from wall surface characteristics (e.g., a smooth wall surface). R mean Small, predicted coating thickness 80μm, Δ Q =0; weld wall surface R mean The predicted coating thickness is 85 μm, Δ Q =5μm). This Δ Q Subject to image feature weights β Influence, β The larger the wall feature, the more significant its influence on the predicted coating thickness. Q The closer the calculation results are to the actual condition of the wall, the more it reflects the priority of quality.

[0034] Δ D To follow α The relevant distance deviation between the drone and the work surface (unit: cm) is the absolute deviation between the drone's actual flight distance and the optimal working distance (the larger the deviation, the worse the paint output from the spray gun, and the more likely it is to cause uneven coating thickness). Δ D The calculation formula is: Δ D =| D - D opt |。The Δ D by α Dominant influenceα The larger the value, the more accurate the actual distance data is for Δ. D The more significant the impact, the more attention will be paid to whether the distance is stable in fusion computing.

[0035] Δ A To follow c The relevant UAV attitude fluctuation deviation (dimensionless, ranging from 0 to 1) is the normalized deviation between the actual rate of change of UAV attitude and the maximum permissible rate of change (the closer the deviation is to 1, the more severe the attitude sway, and the more likely it is to cause paint misalignment; a deviation of 0 indicates a completely stable attitude). Δ A The calculation formula is: .

[0036] Its maximum permissible rate of change is 0.5° / s. This Δ A by c Dominant influence c The larger the value, the better the attitude data is for Δ. A The more significant the impact, the more attention will be paid to the stability of the machine in fusion computing. To obtain the optimal... α , β , c Then, based on the constraints: α + β + c + d =1, the optimal value can be directly derived. d .

[0037] In obtaining P init Simultaneously, compensation is applied to mitigate interference factors affecting the UAV, including gust compensation, load compensation, and pipeline tension compensation. For gust compensation, the condition for detecting gust interference is that the UAV's attitude change rate exceeds a threshold (0.5° / ms). In this case, a gust is detected, and the UAV's attitude offset Δ is collected. i And calculate the gust compensation Δ P wind Δ P wind = k wind ×Δ i . k wind The gust compensation coefficient is used in this embodiment. k wind =0.02MPa / °, adapting to outdoor gust interference range, indicating that 0.02MPa compensation corresponds to every 1° attitude shift. During load compensation, if a change in UAV mass Δ is detected... m If the weight exceeds the threshold (0.5 kg), then the weight is based on the drone's current mass. m real Initial weightm 0. Nozzle spray area 8 S (A general-purpose nozzle with an 8-point spray area can be used) Calculate the load compensation Δ P load Δ P load = k load ×( m 0- m real ) × g / S . k load This is the load compensation coefficient. g This refers to gravitational acceleration. In pipeline tension compensation, if pipeline tension from the drone is detected... F pipe If the threshold (5N) is greater than or equal to the threshold, then based on F pipe Calculate pipeline tension compensation Δ P pipe Δ P pipe =k pipe × F pipe . k pipe The tensile compensation coefficient is set to 0.01 MPa / N to accommodate the range of pipeline tensile interference.

[0038] Finally, according to P init Δ P wall Δ P wind Δ P load Δ P pipe The final spraying pressure was calculated. P final : P final = P init +Δ P wall +0.33·Δ P wind +0.33·Δ P load +0.33·Δ P pipe .

[0039] Furthermore, based on P final It can also be used for vMake fine adjustments. Use the baseline spray pressure. P base Taking 0.8MPa as an example, according to P final Fine-tuning of deviation range v :like P final > P base + Correction value (taken as 0.1 MPa, indicating that the pressure is too high and the paint output is too high), then v Increase the speed by 0.02–0.05 m / s (to increase the speed, reduce the amount of paint output per unit area, and offset the risk of accumulation caused by excessive pressure). If P final < P base - Correction value, then v A decrease of 0.02–0.05 m / s (indicating low pressure and insufficient paint output) means... v Reduce the speed by 0.02–0.05 m / s (slowing down the speed increases the paint output per unit area, offsetting the risk of missed spraying caused by low pressure). If P base - Correction value < P final < P base + Correction value (indicating normal pressure), then v Maintain the same speed and move at the same rate.

[0040] Also according to v The final spraying flow rate was calculated. Q .

[0041] Q = k q × v .

[0042] k q The flow rate correction factor can be based on industry-standard line flow rate experience to ensure a coating thickness of 70-90 μm. Furthermore, a new variable, coating viscosity, can be introduced. or right Q Make corrections: Q = k q × v ×( or ref / or ) .

[0043] In the formula, or ref Reference viscosity (can be taken as) or (The median value). Through or The ratio to the reference viscosity is dynamically adjusted. Q :like or ≥ or ref This indicates that the paint is too thick; at this point, the flow rate should be reduced accordingly to prevent accumulation. If or < or ref The surface coating is relatively thin, so the flow rate is increased accordingly to avoid missed areas and ultimately ensure a uniform coating thickness (70–90 μm). Controllers 1-2 are based on… P final , Q Control the spraying assembly 1-3 to perform the spraying operation.

[0044] based on P final , Q During the wall spraying process, real-time quality closed-loop correction is also performed. This includes: coating thickness correction, attitude fluctuation correction, and PID correction. Coating thickness correction includes calculating the coating thickness based on the image of the wall after spraying. h and calculate h The difference Δ from the target thickness (80 μm) Q real Δ Q real >5μm indicates that the quality deviation exceeds the standard. If Δ Q real > Threshold ( h >85μm), then the Δ in the objective function of the fine-tuning genetic algorithm Q The weight (adjusted from 0.4 to 0.5), and Δ P wall Reduce the pressure by 0.01–0.03 MPa, and reduce the paint output. If Δ Q real <threshold ( h <75μm), then Δ P wall Increase the pressure by 0.01 to 0.03 MPa to make up for the paint output.

[0045] Attitude fluctuation correction includes: the UAV sensors collecting data at a sampling period of 100ms / sample. A If the rate of change of pitch and roll angles remains consistently within the range of 0.3–0.5° / s for 5–10 consecutive sampling periods, it is considered to be fluctuating frequently. This range does not reach the 0.5° / s threshold for gust interference and will not trigger gust compensation; however, even slight aircraft sway may cause spray gun nozzle deviation and uneven coating thickness, requiring targeted correction. The correction method involves adjusting the weighting of the attitude data. c Increase by 0.02~0.03 (as per the original) c=0.25, fine-tuned to 0.27~0.28), the core purpose is to improve the attitude data in the multi-source fusion coefficient. F The importance ratio in the calculation determines the subsequent optimization of weights and calculations in the algorithm. F When the value is set, priority is given to reducing the error caused by attitude fluctuations, and the control of the aircraft stability is strengthened from the data fusion level to avoid the impact of micro-shaking on the coating quality.

[0046] The PID correction targets the regulating valves in the spraying assembly 1-3. These regulating valves are electromagnetic proportional pressure regulating valves, which collect data on the actual spraying pressure in real time. P final deviation e ( t )( e ( t (A positive value indicates that the actual pressure is too high, while a negative value indicates that the actual pressure is too low). Calculate the compensation amount Δ. P PID Overlay P final The driving regulating valve adjusts its opening degree to achieve stable pressure regulation and counteract various interferences. Among them, Δ P PID =Proportional compensation amount + Integral compensation amount + Differential compensation amount. Proportional compensation amount = k p × e ( t Integral compensation amount = k i × e ( t Differential compensation amount = . k p , k i , k d These are the proportional gain coefficient, integral gain coefficient, and differential gain coefficient, respectively. Δ t The sampling period is 0.1s. Δ P PID Overlay P final Gaining new pressure P new The instructions. Through P new The control valve is driven to adjust its opening degree (a positive compensation value increases the opening degree and increases pressure, while a negative value decreases the opening degree and reduces pressure), thus achieving rapid tracking of the actual pressure. P final It can smoothly counteract various disturbances such as gusts, load changes, and pipeline tension.

[0047] In another embodiment, the challenge of spraying paint on irregularly shaped steel structure storage tanks under complex interference lies in the fact that the work object is the surface of a large, curved steel structure storage tank (including weld protrusions and uneven areas), and the operation is subject to intermittent gusts of wind at level 3-4, causing dynamic changes in the tank's mass due to paint consumption. Initial operation parameter settings: D opt =1.6m, P base =0.9Mpa, V base =0.4m / s. When the UAV flies to the weld seam area of ​​the storage tank, the system simultaneously faces multiple coupled interferences: abrupt changes in wall features (weld seam protrusion), attitude fluctuations caused by gusts of wind, and load changes caused by paint consumption. The UAV captures the weld seam protrusion feature in real time through industrial cameras 1-5. The wall feature adaptive quality prediction and compensation algorithm recalculates the feature parameters: mean roughness 1.2μm > 0.8μm, texture complexity 0.8 > 0.6, reflectivity 0.3 < 0.4, and is judged as a first-level working surface. The algorithm immediately outputs the pressure compensation command Δ P wall =-0.15MPa (to prevent paint buildup at raised areas), and simultaneously output speed correction command. v =0.4×(1-0.3×0.8)=0.504m / s, instructing the drone to decelerate and proactively avoid potential defects. Gusts caused the drone's attitude change rate to rise to 0.8° / ms, millimeter-wave radar distance data 1-4 fluctuated to 1.75m, and RTK positioning showed a slight offset. The change in the drone's system state was detected, and the dynamic weights were re-optimized using a genetic algorithm. α =0.35 (increase the weight of distance data) β =0.25 (maintaining image data weights) c =0.25 (increase the weight of attitude data) d =0.15 (location data weight), the fusion decision coefficient is calculated. F =0.65. The algorithm dynamically adjusts the decision priority to trajectory stability > spraying quality optimization, comparing the trajectory deviation Δ S =3.8cm < 5cm, output the adjusted initial pressure command. P init =0.85MPa and distance correction command (reverted to 1.6m). The UAV autonomously identifies interference and calculates compensation. 1. Gust interference: Attitude deviation Δ i =1.2°, substituting into the gust compensation formula: ; 2. Load disturbance: initial total mass m 0 = 19.2 kg, current mass m (t =16.8kg, substituting into the load compensation formula: .

[0048] 3. If the pipeline tension component is ≥5N, calculate the compensation amount using the pipeline tension model. ( The pipeline tension compensation coefficient (taken as 0.01 MPa / N) is included in the final spraying pressure. P final Integrated computing: .

[0049] The output control signal drives the regulating valve to operate, and at the same time outputs attitude correction command (correcting the pitch angle by 1.2°) to ensure that the UAV completes uniform spraying of the weld area on a stable trajectory.

[0050] Post-spraying image quantification results showed an average coating thickness of 78 μm, with a deviation of ≤5 μm, and no defects such as missed spraying or buildup. Following quality data feedback, further parameter fine-tuning was performed: the weighting was optimized to... α =0.32、 β =0.26、 c =0.26、 d =0.16. Anti-interference algorithm corrects gust compensation coefficient. Confirm that the working surface level remains at Level 1.

[0051] In another embodiment, a computer-readable storage medium is also provided, which stores a computer program that, when executed by a processor, implements the steps of the control method for the intelligent outdoor spraying drone described in the above embodiments. This computer-readable storage medium may include, but is not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof.

[0052] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0053] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this invention patent should be determined by the appended claims.

Claims

1. A control method for an intelligent outdoor spraying drone, characterized in that, It includes: Based on images of the wall taken by a drone I The wall surface characteristics obtained by calculation: roughness R mean Texture complexity T complex Reflectivity R ref ; The wall surface is classified into different work surface levels based on its characteristics, and the spraying pressure compensation Δ corresponding to each work surface level is obtained through preset rules. P wall ; according to T complex and the drone's base flight speed V base The corrected speed of the drone was calculated. v ; By comparing the drone's coordinate data with the preset path, the drone's trajectory deviation Δ is obtained. R If Δ R If the value is greater than or equal to the threshold, then the trajectory is corrected; if Δ R If the initial spraying pressure is less than the threshold, then calculate the initial spraying pressure. P init : P init = P base × F ； In the formula, P base As the reference spraying pressure, F To integrate decision coefficients; Compensation for interference factors affecting the drone: If gusts of wind are detected, compensation is made based on the drone's attitude offset Δ. θ Calculate gust compensation Δ P wind If a difference Δ in the mass of the drone is detected... m If it exceeds the threshold, then it is based on the drone's current mass. m real Initial weight m 0. Nozzle spray area S Calculate load compensation Δ P load If the tension in the drone's pipeline is detected F pipe If greater than or equal to the threshold, then based on F pipe Calculate pipeline tension compensation Δ P pipe ; according to P init Δ P wall Δ P wind Δ P load Δ P pipe The final spraying pressure was calculated. P final ; according to v The final spraying flow rate was calculated. Q .

2. The control method for the intelligent outdoor spraying drone according to claim 1, characterized in that, Among the obtained wall features, R mean To extract using the Sobel horizontal gradient method I mean gray gradient Then, it is obtained through linear mapping; T complex To extract edges using Canny edge detection I The number of pixels at the edge of the 5×5 window is obtained; R ref To I The results were obtained after performing grayscale value statistics.

3. The control method for the intelligent outdoor spraying drone according to claim 1, characterized in that, The work surface levels are divided into three levels: Level 1, Level 2, and Level 3. The methods for classifying these levels include: like R mean ≤roughness threshold - ∩ T complex ≤Texture complexity threshold - ∩ R ref If the reflectivity threshold is greater than or equal to 1, it is divided into three levels; If the roughness threshold is < R mean ≤ Roughness threshold 2 ∩ Texture complexity threshold 1 < T complex ≤Texture complexity threshold 2 ∩Reflectance threshold 2≤ R ref If the reflectivity threshold is less than 1, it is classified into two levels; like R mean > Roughness threshold two ∩ T complex >Texture complexity threshold two∩ R ref If the reflectivity threshold is less than 2, it is classified as Level 1.

4. The control method for the intelligent outdoor spraying drone according to claim 1, characterized in that, v The calculation method is as follows: v = V base ×(1- k v × T complex )； In the formula, k v The correction factor is derived from industry experience. And / or, if P final > P base +correction value, then v Increase by 0.02–0.05 m / s; like P final < P base - Correction value, then v Decrease by 0.02–0.05 m / s; like P base - Correction value < P final < P base +correction value, then v constant.

5. The control method for the intelligent outdoor spraying drone according to claim 1, characterized in that, Δ R The method is as follows: The actual trajectory is obtained by fitting the three-dimensional coordinate sequence of the UAV; the deviation value obtained by comparing the actual trajectory with the trajectory of the preset path is Δ. R ; F The calculation method is as follows: F = α · D norm + β · I feature +γ · A stab + δ · L acc ； ； In the formula, α For distance data weights, β Image feature weights, γ As the weights for the attitude data, δ To determine data weights, D norm The optimal operating distance for drones. I feature for I The comprehensive image feature values, A stab Let be the rate of change of attitude of the drone. L acc This represents the positioning error.

6. The control method for the intelligent outdoor spraying drone according to claim 5, characterized in that, Genetic algorithm optimization α , β , γ , δ The optimization objective is to make the weighted sum of the comprehensive errors... J Minimum, the constraint is: α + β + γ + δ =1; where the optimization objective function is: minutes J= 0.4D Q+ 0.3D D+ 0.3D A ; In the formula, Δ Q To follow β Related coating quality deviation, Δ D To follow α The relevant distance deviation between the drone and the work area, Δ A To follow γ Related drone attitude fluctuation deviations.

7. The control method for the intelligent outdoor spraying drone according to claim 6, characterized in that, Control methods also include: based on P final , Q After spraying the wall, dynamic adjustments are made based on feedback on the spraying quality. P final Dynamic correction P final The methods include: The coating thickness was calculated based on the image of the wall surface after spraying. h and calculate h The difference Δ from the target thickness of the process Q real ; If Δ Q real If the threshold is greater than Δ, then increase Δ. Q The weights, and Δ P wall Reduce by 0.01–0.03 MPa; If Δ Q real < threshold, then Δ P wall Increase by 0.01 to 0.03 MPa.

8. The control method for the intelligent outdoor spraying drone according to claim 1, characterized in that, During the process of compensating for interference factors of drones, the condition for detecting gust interference is: the drone's attitude change rate is greater than the threshold. Δ P wind The calculation formula is: Δ P wind = k wind ×Δ θ ; k wind This is the gust compensation coefficient; And / or, Δ P load The calculation formula is: Δ P load = k load ×( m 0- m real ) × g / S ; k load This is the load compensation coefficient; g It is the acceleration due to gravity; And / or, Δ P pipe The calculation formula is: Δ P pipe =k pipe × F pipe ; k pipe This is the tensile force compensation coefficient; And / or, P final The calculation formula is: P final = P init ×Δ P wall +0.33·Δ P wind +0.33·Δ P load +0.33·Δ P pipe ; And / or, Q The calculation formula is: Q = k q × v ; k q This is the flow correction factor.

9. An intelligent outdoor spraying drone, characterized in that, It includes: Organism; Spraying assembly, used for spraying paint onto walls; The controller uses the control method of the intelligent outdoor spraying drone as described in any one of claims 1 to 8 to control the spraying pressure and spraying flow of the spraying assembly.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of the control method for an intelligent outdoor spraying drone as described in any one of claims 1 to 8.

Images