Abstract:

For a class of smooth nonlinear multivariable systems whose working-points vary with time, a state-dependent auto-regressive with exogenous (SD-ARX) model and its functional coefficients are composed of the Gaussian radial basis function (RBF) networks with some state variables representing the system’s nonlinear dynamic characteristics. The model is called a state-dependent RBF-ARX model and estimated by a structured nonlinear parameter optimization method (SNPOM) offline. The nonlinear predictive strategy is designed based on the state-dependent RBF-ARX model that does not require online parameter estimation so as to improve the real-time performance of control systems greatly and has a preferably control performance. A case study on a simulator of a quadrotor illustrates the effectiveness of the nonlinear modeling and the feasibility of the control method.