Adaptive Control Algorithms

The thinking behind adaptive control is to combine the processes of system identification, to gain a system model for an unknown plant, and of feedback control design for a known plant model. This yields a design-while-identify control strategy otherwise known as self-tuning or self-adjusting. To be naive one would suppose that a good adaptive control procedure is able to be manufactured by cobbling together a good parameter identifier and a good control law, and all could be expected to perform. This is partially true in that a successful adaptive control law must incorporate both of these features, but this may not be sufficient for good performance. A third feature needs to be taken into account which addresses the marrying of the identification objective with the control objective through the design or selection of adaptation speed, model structure, signal conditioning, control design rule, reference signal properties, speed of plant variation etc.The aim of this work is to present and develop a view of the salient aspects of adaptive control algorithms with the intent of encapsulating their dominant features and explaining the reasons for their specific numerically desirable properties. We shall concentrate upon the development of a coherent standpoint from which the practical aspects of adaptive controller design emerge.