基因演算法應用於二階多變數內建模式控制之研究

Design of Digital Two-step MIMO Internal Model Control by Genetic Algorithms

內建模式控制已被證實是一項實用且有效的控制器，研究人員已體會到內建模式控制特別適用於時間延遲系統。傳統上，設計內建模式控制之前必須先求取控制系統轉移函數，然而轉移函數經數位化處理，往往牽涉到反轉之後的轉移函數是否穩定，控制訊號是否發散與飽和等問題。本文提出以一階時間延遲模式簡化多變數控制系統轉移函數，並以嶄新的數位二階控制法則作為設計多變數內建模式控制的依據。為有效驗證二階多變數內建模式控制，本文藉由基因調準法則針對高階長時間延遲的轉移函數進行模擬，並與基因調準比例－－積分控制的結果作一比較。結果證實基因調準二階多變數內建模式控制的輸出響應大幅提昇26.57%，因此被認定是一項有效且實用之控制器。

Having been demonstrated to be an effective control for MIMO time-delayed plants, the single-step multivariable internal model control will then be extended to the design of the two-step multivariable internal model control. To facilitate to measure output performance, digital control is used in the paper. Summation of squared error between output and the step input at each sampling time is used to measure the output performance. The genetic tuning methodology will be deployed to automate the tuning of the two-step MIMO internal model control. Also MIMO robustness analysis using genetic algorithms for measuring the input real perturbation gain margin and output perturbation phase margin will be implemented in this paper. To gain the clear insight on the output performance of the two-step MIMO internal model control, the result will be illustrated by an example of a MIMO plant with time delay, and then comparing with the centralized PI control.