Study On The Deposition Distribution And Evaluation Method Of Spray Droplets For Plant Protection UAV

Pests and diseases are the main factors that hinder the crops’growth and limit food production.Chemical control is currently the main initiative to solve the pest and disease problems.Plant protection UAV has gradually replaced traditional application tools and has been widely used with its advantages of high efficiency,flexibility,and environmental adaptability.However,due to the lack of clarity about the atomization principle,the movement mechanism of droplets and the interaction between the liquid and the target leaves,there are problems such as drift and loss of liquid,which not only endanger the ecological environment but even threaten the safety of humans and animals.Therefore,a comprehensive understanding of the spatial distribution and deposition effects of droplets and the establishment of a reliable predictive model are the keys to improving pesticide utilization during the plant protection spray.In this study,we focus on the whole process of plant protection UAV spray.The three-step research path of"UAV-Droplet-Leaf"was proposed,the rotor wind field distribution law of the quadrotor UAV was explored,the prediction model of droplet size distribution was established,and the effects of droplet properties and surface properties on the static and dynamic interaction results of droplets on the leaf surface were analyzed.Numerical simulation was carried out by computational fluid dynamics（CFD）to achieve the visual analysis of dynamic processes,such as wind field distribution of UAV,droplet deposition distribution,and droplet impact on leaves.It is important for guiding the setting of spraying parameters and the adjustment of spraying modes.The main research contents and conclusions are as follows:（1）Aiming at the problem that the continuous distribution of the rotor wind field was unclear and the numerical simulation analysis was less,the rotor wind field distribution of the quadrotor UAV was explored,and a simulation model of the rotor wind field distribution was established.The horizontal distribution of the rotor wind field at different heights was measured by a three-way wind speed sensor network system,and the simulation effects of different turbulence models were compared.The spatial distribution of the rotor wind field was visualized.The results showed that:（1）the Y-direction wind speed was the main wind speed of the rotor wind field,and the Y-direction wind speed showed a trend of decreasing and then increasing with the increasing distance from the center position;（2）the SST6)-model had the best simulation effect by comparing with the measured wind speed at the test positions which had a good measurement accuracy.At the center of the rotor wind field at each test height,the SST6)-model had the smallest error rate and the total error rate was only 11.458%;（3）the hoisted airflow appeared at the area close to the edge of the rotor.With the increase of the rotor height,the Y-direction wind speed increased first and then decreased.The X-direction wind speed value of xoy plane and Z-direction wind speed value of zoy plane first increased and then gradually decayed to a lower level.Due to the rotational direction of the rotor,the phenomenon of"inflow and outflow"led to the difference in the wind field width in different planes.For the xoz plane of different heights,the maximum wind speed of the Y direction decreased,and the wind field width increased with increasing height.The high-speed area gradually converged and finally formed an elliptical wind field distribution area.（2）Aiming at the problem of lack of comprehensive consideration of multiple factors and poor prediction effect of droplet size distribution,the effects of spray parameters and spatial parameters on droplet size distribution were analyzed,and quantitative analysis models of droplet size distribution were established based on machine learning methods.The spray test system was used to measure the droplet size distribution of different flow rates,nozzles and spatial locations.The prediction models of volume median diameter（VMD）and relative span（RS）were established based on multi-layer perceptron（MLP）,extreme learning machine（ELM）,decision tree（DT）,support vector regression（SVR）,radial basis function neural network（RBFNN）and polynomial regression.The results showed that:（1）the increase of nozzle orifice diameter and the decrease of flow rate both increased the VMD and decreased the RS;（2）the VMD and RS curves were centrosymmetric in spatial distribution.With the increase in spray height and horizontal distance,VMD increased;and there was an obvious inflection point on the horizontal position of the RS curve,within which RS decreased significantly with increasing horizontal distance and decreasing height.When the horizontal distance was greater than this position,the influence of spatial parameters on RS was not obvious;（3）The VMD and RS prediction models based on machine learning achieved better results than polynomial regression.Among them,the SVR model had the best prediction effect for the prediction of VMD,of which correlction coefficient of prediction（R_p）and root mean square error of prediction（RMSEP）were0.9929 and 6.0690,respectively.For the prediction of RS,the MLP model had the best prediction effect,of which R_p and RMSEP were 0.9537 and 0.0398,respectively.Compared with the machine learning methods,the polynomial regression model had a larger error in predicting RS,and its R_p was only 0.5989.（3）Aiming at the problem that the tests of droplet deposition distribution were limited by field experiments and wind tunnel experiments,the deposition distribution and drift laws of spray droplets were analyzed based on numerical simulation.Based on the discrete phase model（DPM）,the spray process of the hollow cone nozzle was simulated,and the deposition distribution and drift of spray droplets were analyzed under different wind speeds and deposition heights.The results showed that:（1）the spray simulation effect of TR80-02C and TR80-03C nozzles was better,and the relative errors were less than 10%;（2）when the deposition height was greater than 0.5 m,the peak value of deposition first increased and then decreased with the increase of wind speed.When the deposition height was less than 0.5 m or the wind speed value was less than 0.5 m/s,the effective deposition rate of droplets reached more than 95%.The higher the deposition surface was,the higher the wind speed was,and the lower the effective deposition rate of droplets was;（3）with the influence of the wind speed,spray droplets appeared stratified in space.Small droplets were more likely to drift and be located downwind.The higher the wind speed was,the more obvious the stratification phenomenon was.As the wind speed increased,the droplet drift amount and the droplet size distribution(D₁₀,VMD and D₉₀)increased rapidly and then flattened out.（4）Aiming at the problem of incomplete consideration of the influencing factors on leaf surface wettability and unclear interaction mechanism,the effects of droplets properties and physicochemical properties of leaf surfaces on wettability were investigated.Taken the rape as the research object,the influence of different pesticide types,droplet size and the adding of adjuvants on wettability was investigated by contact angle measurement,and the influence of surface properties on wettability was evaluated by surface free energy（SFE）and surface microstructure.The results showed that:（1）the contact angles of different test liquids on the rape leaf surface were positively correlated with their surface tension and the correlation coefficient was 0.941;（2）the size of the spray droplets had no significant influence on the wettability within7μL in size;（3）the addition of organosilicon adjuvant resulted in a better improvement for pesticides with poor wettability on rape leaf surfaces;（4）the rape leaf surface had a low SFE,of which the dispersion component took up a large proportion,and the complex microstructure made the surface rough so that the surface was hydrophobic;（5）Aiming at the problem that the impact mechanism between droplets and the leaf surface was not clear and the impact results were difficult to predict,the effects of different characteristics of droplets and surface structures on the impact results and spreading characteristics were analyzed,and the process of droplets impacting the leaf surface was restored based on numerical simulation.Based on the droplet impact analysis system,the impact results and spreading of different droplets on different surfaces（rape,rice,pepper,wax and water-sensitive paper）were observed,and the process of droplets impacting the rape leaf surface was simulated.The results showed that:（1）for the hydrophobic leaf surface,the droplets were more likely to bounce or shatter on the adaxial surface of rice with poor wettability.When the droplet impacted the hydrophilic pepper surface,there was no bounce,and the droplet would shatter directly after the Weber number increased;（2）rice and pepper leaves were more sensitive to droplets.When the droplet diameter increased,the beginning and ending Weber numbers of rice leaves would decrease slightly in the"bounce-shatter"transition region,while those of the"adhesion-shatter"transition region on pepper leaves increased significantly with the droplets;（3）the addition of the adjuvant increased the wettability and the adhesion probability of droplets on rape leaves.The key Weber number in the transition region of"bounce-shatter"had a high correlation with the adjuvant concentration.The determination coefficient of the linear fitting between the beginning and ending Weber numbers with the concentration reached 0.9386 and0.9604,respectively;（4）the impact characteristics of droplets on rape leaf surface obtained by numerical simulation had a high consistency with the actual observation results,which provided important reference value for analyzing the interaction between droplets and leaves.