Modified pole-assignment controller for plant models with exact or near pole-zero cancellation

A design technique which uses a weighted least-squares approach for the solution of the output feedback pole-assignment problem is presented. The technique may be used to assign some closed loop poles to positions which improve the conditioning of the pole-assignment algorithm. The link between ill-conditioning and the disturbance response of the system is demonstrated. The least-squares approach is also applied to the design of non-minimal order controllers, in order to reduce input variance. Examples are given, comparing the noise performance obtained from such designs with that obtained using non-minimal controllers designed explicitly to minimise variances due to noise. The use of the technique, to select some closed loop poles in an adaptive control system where the order of the plant model has been overspecified, is demonstrated.