Offset Free Model Predictive Control For A3-DOF Helicopter

The3-DOF helicopter experimental platform is a typicalmulti-input/multiple-output, nonlinear, strong coupling system, which hasrecently become a powerful auxiliary platform for teaching andresearching of control theory. Due to input and output constraints, modelpredictive control algorithm is a good candidate of control algorithms forit. Based on analyzing the dynamic model of the Googol3-DOFhelicopter system, the paper points out that the existing linearized modelhas error in the direction of gravity, which may cause steady-error whentracking desired orbit.The main research work of this dissertation can be summarized asthe following three aspects:1) According to the characteristics of the3-DOF helicopter model,combined with offset-free predictive control method, an offset-freemodel predictive controller with feedforward is designed. AndSimulink simulation platform verify the effectiveness of the controlalgorithm, The simulation results show that the proposed controller isable to track the reference trajectory well with no steady-state error,meanwhile also maintain good robustness.2) Model predictive controller requires solving quadratic programmingproblems, According to Wolfe-simplex method, we developed aquadratic programming solver, realizing complementary slacknessconditions, tested it on benchmark, and compared the results with thatcalculated by MATLAB.3) The mismatch between dynamic model and the real plant affects the control accuracy. For real experiment, the model parameters arere-identified. Based on the identified model, both unconstrained andconstrained MPC are implemented for a3-DOF helicopter real timecontrol. Tracking experiments for single axis and two axis are made,which show good control performance.