Suboptimal adaptive control of a class of non-linear systems†

The problem of controlling a class of nonlinear systems with uncertain parameters and imprecisely defined system and measurement noise terms is considered. A sub-optimal adaptive control algorithm is developed as a solution to coupled linear modelling and control optimization problems. Features of the algorithm are the forward-recursive generation of the modelling and control processes and necessity for past and present values of states and parameters alone in order to implement the algorithm (i.e. no precomputed nominal trajectory is required). These features make the algorithm ideally suited to real-time control applications Real-time control of a non-trivial fifth-order system, (a model of a plasma torch furnace) is considered and the results of an experimental study are presented.