Title :
Comparison of CMAC controller weight update laws
Author :
Kraft, L.G. ; An, Edgar ; Campagna, D.P.
Author_Institution :
Dept. of Electr. & Comput. Eng., New Hampshire Univ., Durham, NH, USA
Abstract :
A modeling technique that allows direct analysis of stability and convergence properties for control systems using the cerebellar model articulation controller (CMAC) neural network approach is presented. Two different network weight-updating methods are modeled and compared. The first technique updates the weights after each training sequence. The second method updates sequentially during each control cycle. Results favor sequential updating. In both weight methods the CMAC method can be made unstable
Keywords :
brain models; control system analysis; neural nets; stability; CMAC controller weight update laws; cerebellar model articulation controller; convergence; neural network; stability; Control system synthesis; Control systems; Convergence; Eigenvalues and eigenfunctions; Equations; Large-scale systems; Matrices; Neural networks; Stability analysis; Weight control;
Conference_Titel :
Decision and Control, 1989., Proceedings of the 28th IEEE Conference on
Conference_Location :
Tampa, FL
DOI :
10.1109/CDC.1989.70451
