進化型計算を用いた固定構造動的量子化器の設計

抄録

This paper addresses the design problem of fixed-structure dynamic quantizers for discrete-valued input control. The problem is to find a dynamic quantizer when the dimension and the decentralized structure of quantizer are given as design specifications. In this paper, we first formulate a quantizer design problem under constraint conditions such as stability and input saturation. Then, for getting a solution to the problem, we provide a simple and useful design method with a particle swarm optimization which is one of evolutionary algorithms. Finally, we verify that the proposed method gives satisfactory fixed-structure dynamic quantizers, by several numerical examples.