Design of Nonovershoot MRACS With Application to D.C. Servo Motor System

Excessive overshoots in a transient response are undesirable in a model reference adaptive control system (MRACS) and have to be avoided in practical applications. This paper discusses the design of an MRACS with no overshoot. In this design a d-step ahead estimator is introduced to evaluate the expected maximum and minimum values of the plant output. According to these estimates, the adaptive control input is adjusted so that the output has no overshoot. For the estimator and the input adjustment an existence region of the unknown plant parameters must be known. It is obvious that the smaller the existence region is, the better is the estimation and adjustment, and, consequently, the MRACS performance. First, an algorithm which successively reduces the region is presented. An initial polyhedron region V(0), which includes the unknown parameters, assumed to be given. The volume of V(k) containing the unknown parameters is then successively reduced by a projection-type algorithm which uses the input and output of the plant. Next, the design of an MRACS is discussed in which this region V(k) plays an important role. The proposed controller generates an adaptive control input sequence which makes the plant output follow the reference output without any overshoot. The proposed MRACS was applied to the adaptive position control of a D.C. servo motor system with an unknown load. Experimental results demonstrate the usefulness of the proposed design.