On Tuning Regulators

Up to this point in the book, we have assumed that the plant model has been given, and the objective is to carry out system analysis and controller design. In many situation, however, a good model is not available. The two typical solutions to this problem are (i) to model the plant and identify the parameters and (ii) use an adaptive controller. In both cases certain plant sructural information is typically needed, such as the plant order, which may be available in simple cases but not readily so in more complicated situations. In this chapter we examine a more direct approach, wherein the goal is to design a controller for the plant without the intermediate step of obtaining a detailed model; it is a generalization of the classical approach to tuning a three-term PID controller used for single-input single-output systems. The method requires the plant to be stable and it assumes that (i) a finite number of experiments can be carried out on the plant, and (ii) it is possible to carry out on-line tuning of the controller parameters. While the method is presented for the case of step tracking, the approach can be generalized.