Abstract
This paper addresses the problem of controlling linear, time-invariant deterministic systems subject to plant parameter uncertainties and fewer measurable outputs than states. A dynamic controller is developed, whose structure is specifically that of a Luenberger state observer (Newmann 1970, Goldstein 1972). A minimax control formulation is postulated, such that system disturbances resulting from parameter misalignments are modelled as extremum disturbances. The optimization procedure results in a set of non-linear algebraic matrix equations which, when solved recursively, determine a sub-optimum set of feedback gains and state observer parameters. Application of the design algorithm to a single-output, second-order system demonstrates that convergence to a unique minimum results.

This publication has 5 references indexed in Scilit: