Development of an Adaptive Control Toolbox

The subject of adaptive control has been studied for over forty years. In spite of this, very few practical applications of adaptive control have been achieved. There are several possible reasons for this. Firstly there are many identification and control schemes to choose from, none of which are simple to tune to work in real world applications. Secondly, the algorithms involved are non trivial and it is only in recent years that suitable computer platforms have existed for their implementation. Lastly, because of the non trivial nature of the algorithms they are difficult to code. These problems in the practical application of adaptive control highlight the need for a real time adaptive control 'toolbox' running on a fast computer platform. Such a toolbox should implement a range of identification and control algorithms together with all the associated filtering, saturation limits, dead zones, etc. that are required for the practical implementation of the algorithms. This has been achieved at the University of Newcastle, and is called UNAC for 'University of Newcastle Adaptive Controller'. This paper gives an overview of the capabilities of UNAC. Fundamental to UNAC is the implementation of algorithms using a new discrete time operator, the delta operator, in place of the more usual shift operator q. The use of this operator provides much greater numerical robustness in the implementation of algorithms thus allowing higher sampling rates relative to the plant bandwidth to be used. It is also heuristically more appealing by allowing continuous time insights to be applied to digital control problems. A discussion of this new operator and the advantages of its use are given.