1994 Volume 7 Issue 3 Pages 77-83
A generalized truncated least squares adaptive algorithm is presented in this paper. The algorithm is a generalization of the algorithms by the truncated least squares method (TLSM) and the least squares method (LSM). The TLSM utilizes truncated data and is effective for systems with drastic parameter variations. This method is sensitive for small parameter variations and also for any disturbance. To accomodate the variations of the estimated parameters, a performance index of TLSM is generalized to include a new term of parameter variations in the form of a linear combination. The combination parameter introduced acts as an ability of adaptivity.
Adaptive control experimentations of the liquid level controlled system is presented which show the effectiveness of the proposed algorithm.