Research On Flight Control Algorithm Of Multi-Rotor UAV

In these years,the agricultural plant protection unmanned aerial vehicles become a research hotspot.As this field has some requirement of UAV flight performance,such as flight stability,reaction timeliness and flight control system compatibility,the flight control algorithm of six rotor unmanned aerial vehicle(UAV)is experimented and studied in thesis report.Firstly,the structure and dynamics characteristics of the control object are analyzed,which lays the foundation for the establishment of the simulation model and control system.In addition,in order to avoid the attitude variables interfere with each other,the report improves and optimizes the control method,and the adaptive adjustment of the control parameters is realized.The mathematical model can't only reflect the relationship between the variables,but also show the characteristics of attitude change.Therefore,the thesis report studies and analyzes the propeller model,kinematics model and control allocation model,with the kinematic characteristics of the six rotor UAV and the performances of its main components.In order to visually view the movement of control object,the machinery model of agricultural plant protection UAV can be built in the Adams software,then executing joint simulation with MATLAB.In order to create a concise model,the simulation model of UAV can be built in the MATLAB software,by 6DOF and other Simulink function blocks.The control allocation model is an intermediary transformation part between the controller's output values and the control object's input values.This part has big difference between UAVs of different rotor number.So,it's necessary for its analysis and constructionIn addition,the traditional PID method is optimized.The parameters of PID method are difficult to be optimized and the system adaptability is poor.But neural network has the characteristics of robustness,nonlinear approximation and autonomous learning.Six rotor UAV is non-linear underactuated and high coupling system,and it has complex mechanical structure without a uniform mass.So,this report combines BP neural network and PID method together,and calls this control strategy PIDNN(PID neural network)method.This method does not use the PID method explicitly,but sets proportional neurons,integral neurons and differential neurons in the hidden layer of neural network,and the connection weights are optimized in each iteration.This report compares the simulation results of PIDNN controller with results of PID controller,verifying that PIDNN method accelerates the speed of the six-rotor UAV to get to the target state,decreases overshoots,and reduces the number of manual adjustments.So,PIDNN controller is more valid in reducing adjustment time and overshoots.