Identification of the pure time delay of a process

Previous work by the author has produced a novel scheme for controlling processes with pure time delay, Mitchell (1987, 1990). This scheme uses a model of the minimum phase elements of the process to generate the necessary control signals. This model must provide an accurate representation of the process, and so must be adaptive and identified continually. The identification method used is the recursive least squares algorithm and this produces a suitable model. However, the method requires that the value of the time delay of the process be known. Usually this is assumed as a priori information, but even if its value is known, it may change during the operation of the process. In the paper various methods are considered for identifying the time delay, and then a new method developed by the author is described and its performance evaluated