I just developed a general solution to multivariable decoupling control which is optimal and analytical. It should be, but not confirmed, a breakthrough. Simple testable design examples are provided at http://automation.sjtu.edu.cn/wdzhang/contribution.htm . Comments or suggestions from researchers in this field or interested readers are highly welcome.

Email: wdzhang@sjtu.edu.cn

Brief statement of the problem

Given a linear plant, analytically design the optimal and suboptimal controller such that 1) The closed-loop system is internally stable; 2) The closed-loop response is decoupled; 3) The performance and robustness can be quantitatively tuned.

The plant can have right half plane zeros, right half plane poles, and time delays.

Features

Compared with the classical decoupling control, the proposed method has two important merits: 1) It is applicable to unstable plants. 2) The robustness issue can be naturally treated by those developed techniques.

What is more, the proposed method exhibits some other attractive characteristics: 1) It can be used for the control of unstable NMP plants with time delay. 2) The closed-loop response is decoupled and the design procedure is optimal and analytic. 3) It provides simple tuning rule for quantitative performance and robustness. 4) No state variables and no observer are used. 5) No augmented plant is used and the controller is of lower order. 6) The method can directly deal with integrating plants. 7) It can be used for the design of multivariable PID controllers.

-- www (www@yahoo.com), December 09, 2004