Design And Experiment Of Variable Target Spray System For Plant Protection UAV

In the rice planting process,due to human factors such as uneven planting and unreasonable fertilization,or natural disasters such as insect pests,typhoons,and floods,plant necrosis.The rice coverage in some areas is very low.In the process of plant protection UAV operation,if the area is applied with chemicals,it will cause pesticide waste.Besides,the traditional plant protection UAV will cause uneven spray and low utilization rate of pesticide due to the change of flight speed during start braking and steering.To improve the above problems,this paper designs an adaptive variable target spray system for plant protection UAV.The system can collect data such as flight speed,field image,and spray flow in the process of UAV operation and perform adaptive variable spraying according to the flight speed,and calculate crop coverage of field images to control the spray state of corresponding sprinklers to achieve target spray.The system takes Raspberry Pi 3B+ as the core processor,and the main components include the camera,wind pressure transmitter,flow sensor,micro diaphragm pump,solenoid valve and,electric centrifugal sprinkler.The functions of the system include information collection,information analysis,information processing,spray decision-making,and spray control.In the design process of the system,the relationship between PWM the duty cycle and spray flow rate at PWM control signal frequencies of 3Hz,6Hz,9Hz,and 12 Hz is measured,and the relationship between 5%～80% duty cycle and the actual spray flow rate is established according to the system requirements.BP neural network PID control algorithm is constructed,and MATLAB is used to simulate step response,sine tracking,and square wave tracking of PID,fuzzy PID,neuron PID,and BP neural network PID respectively.Simulation results show that the BP neural network PID control algorithm has the best control effect,and it can better meet the real-time and stability of the UAV adaptive variable spray system.The rice images are collected,and their coverage is calculated by manual segmentation.The gray-scale method of super green method,super green and super red method,and standard difference index method is combined with the gray-scale mean value,OTSU,and maximum entropy value of image segmentation algorithm,respectively.The comparative analysis of the relative error and time-consuming of the coverage shows that the super green and super red maximum entropy method takes less time and has less error.After completing the design of the system,the system was mounted on the drone,and the target spray and droplet deposition tests were carried out separately.The variable spray test results show that the system can control the spray flow according to the speed of UAV,the average deviation between the actual flow and the target flow is 8.43% and can improve the problem of uneven spray due to flying speed.The target spray test results show that droplet deposition,droplet number and droplet coverage rate decrease in full spray,semi spray and non spray areas,and the system can reduce pesticide usage.The results of the droplet deposition test indicate that the primary and secondary order of the factors affecting the density of droplet deposition is flight height,wind direction,nozzle voltage,spray pressure,flight speed,and wind direction.The primary and secondary order that affects the spray uniformity is flight speed,wind direction,flight altitude,wind speed,spray pressure,and nozzle voltage.The system built in this paper can effectively improve spray uniformity and improve pesticide utilization to a certain extent,provide reference directions and decision support for the application of agricultural aviation variable spray and target spray technology,and help the development of plant protection drones.