抄録
The trajectory sensitivity method, which utilizes a quadratic performance index with the trajectory sensitivity term, is well known as an approach to design a regulator explicitly accounting plant parameter variations. However, most existing design methods are based on heuristic iterative algorithms that can not be guaranteed to converge. In this paper, for discrete-time systems, we formulate the trajectory sensitivity method as a parametric LQ problem. To solve the problem, we first propose an algorithm based on the iterative solution of two Lyapunov equations. Although this algorithm is practically efficient, its convergence can not be guaranteed theoretically. Then we propose a modification to guarantee the convergence. As an illustrative example, we give a design example for a simple second order system with an uncertain parameter.
