Gain Scheduling Control System Design Based on Numerical Optimization via Matrix Expansion and Sum of Squares

点击翻译按钮获取中文摘要

Firstly, we confirmed that our proposed GS (gain scheduling) control based on SOS (sum of squares) had good performance compared with the SDRE (state-dependent Riccati equation) method through some experiments applied to an ADIP (arm driven inverted pendulum). Secondly, we investigated a GS control system design whose parameter variation region is not a box but a superellipse or two quadratic curves. As a result, performance improvements were illustrated by non-box regions. Thirdly, we investigated the computational methods of estimating the DA (domain of attraction) for the descriptor non-polynomial system. Finally, the DE (Differential Evolution) method were used to get an approximate solution of the nonlinear optimal regulator problem. In proposed method based on the DE, we could design a simple GS controller.