Experiment Platform Design And System Identification Research Of Quadrotor

Quadrotor helicopter is a multi-input and multi-output sytem whose nonlinear characteristics make it very difficult to obtain its dynamic model. However, it can be relatively easy to obtain the mathematical model of the aircraft through system identification. Test signal design is a key element to achieve the success of the final identification. Optimal test signal is designed to get the system’s largest process information using minimum test signal power. This work includes experimental aircraft platform and closed-loop optimal testing signal design, identification of aircraft with optimal testing signal, optimization of PID controller parameters based on the identification model, and validation of feasibility of the proposed identification method by simulation and physical experiment. The main contents are outlined as follows:1. The hardware of experimental platform design is completed. Detailed analysis of quadrotor’s principle and mechanism model is done and this work show that quadrotor complete the six degrees of freedom motion by the motor speed difference under the "X" mode. Law of quadrotor’s motion and control feature via closed-loop simulation is grasped; The whole hardware of flight control system based on the feedback control framework is designed; Hardware selection and circuit design is completed.2. Software program design according to the hardware experimental platform is completed. Control algorithm is implemented based on cascade PID control: inner loop control for three-axis angular rate, outer loop control for attitude angle; An quaternion attitude algorithm based on the complementary filter is proposed combining different attitude measurements’ complementary characteristics of high-low frequency, and the algorithm can solve the problem of "Gimbal Lock"; low-level control and PC software code are written according to the design goals. Software module and flight testing of experimental platform are completed.3. Optimal closed-loop testing signal design method for maximum information is proposed. This method will expand parameter covariance matrix in the frequency domain to construct a convex optimization problem whose constraint is linear inequalities with full consideration of the noise model; Test signal is changed into a cosine signal superimposed expression; The design method is verified via Monte Carlo test which indicates that under the same level of test energy, optimal signal can get a more accurate model than white noise signal and generalized binary noise signal.4. Quadrotor’s closed-loop identification experiment is done. The simulation with application of PEM identification method and optimal test signal is done. The simulation results validate that quadrotor system identification can obtain an accurate model based on the optimal testing signal; Closed-loop identification and optimization of PID controller based on identification model is done with the use of flight experimental platform; Experiment result validates that the control performance can be improved by this method.