Adaptive fuzzy control for flexible-link manipulators: A hybrid frequency-time domain scheme

The authors address the implementation of an adaptive fuzzy controller for flexible link robot arms. The design technique is a hybrid scheme involving both frequency- and time-domain techniques. The eigenvalues of the open-loop plant can be estimated through application of a frequency-domain-based identification algorithm. The region of the eigenvalue space, within which the system operates, is partitioned into fuzzy cells. Membership functions are assigned to the fuzzy sets of the eigenvalue universe of discourse. The degree of uncertainty on the estimated eigenvalues is expressed through these membership functions. The knowledge database consists of feedback gains required to place the closed-loop poles at predefined locations. A rule-based controller infers the control input variable, weighting each gain with the value of the membership functions at the identified eigenvalue. The controller was demonstrated in simulation studies.<
>