1991 年 27 巻 5 号 p. 538-545
In this paper we propose a design method for the weighted minimum variance self-tuning controller with reduced output variance. This is achieved by introduction of a polynomial that gives freedom to controller parameters.
However, calculation of the controller parameters is complex and requires excessive time. To simplify the calculation, two methods are proposed in this paper.
The first method is to choose a certain polynomial as a closed-loop characteristic one. In this case the output error becomes moving average of white noise. Calculation is simplified and the effect of reducing output variance is found.
Another method is an approximation of the optimal solution. It is achieved by disregarding the effect of closed-loop characteristic polynomial on output variance.
The two methods reduce effectively the output variance and simplify the calculation.
Finally, the result of computer simulation of the adaptive control is shown. A linear time-in-variant unstable nonminimum phase plant and a linear time-varying plant are used to illustrate the effectiveness of the proposed method.