Optimal moving average filter with application to self-tuning type control against deterministic disturbances

Abstract

An optimal moving average (OMA) filter for a linear stochastic controlled plant is developed from the viewpoint of the internal model of the deterministic disturbance. The OMA filter enables the minimum-variance self-tuning control not only to overcome effectively various kinds of deterministic disturbances modelled by the linear homogeneous difference equation (such as stepwise and sinusoidal disturbances), but also to depress greatly the amplification of random noise in the control system. Thus, both the precision of parameter estimation and the quality of control can be improved, especially in the cases of strong coloured noise. The basic idea of the OMA filter is also extended to the multivariable system case. Several numerical examples have shown the effectiveness of the OMA filter in different cases. Some remarks on the use of the OMA in self-tuning control are also presented.