分片线性逼近及其在预测控制中应用.pdfVIP

AHH AHH AHH TP A Piecewise Linear Approximation and Its Application in Model Predictive Control XU Jun, HUANG Xiaolin, WANG Shuning (Department of Automation, Tsinghua University, Beijing, China 100084) Abstract: The piecewise linear approximation is introduced in this paper. The approximation of nonlinear functions by piecewise linear functions may simplify the nonlinear system via linear methods. The choice of a good piecewise linear model is vital in the application. Therefore, three kinds of compact piecewise linear models are introduced, of which the model of adaptive hinging hyperplanes (AHH) is used in model predictive control (MPC). The nonlinear system to be controlled is modeled via AHH approximation, and the modeling is simple and efficient due to the adaptiveness of AHH identification algorithm. After the predictive model is built, an open-loop optimization problem is formulated to get the optimal control sequence and the receding horizon control strategy is employed. The open-loop optimization problem is actually a series of convex optimization subproblems and the existence of the global optimum is guaranteed for the original problem. In order to make an efficient application, a decent algorithm is proposed to search for a local optimum. Simulation results show the prospect of the AHH-based