Special Issue "Fuzzy/Neural Network Applications to Dynamics and Control of Mechanical Systems". Use of Neural Networks as Numeric-Symbolic Converters for Gain Scheduling Adaptive Control Systems.
Goro Obinata, Yasushi MURAGISHI · TRANSACTIONS OF THE JAPAN SOCIETY OF MECHANICAL ENGINEERS Series C · 1992
A new control scheme is proposed in this paper which can cope with environmental variations resulting from load disturbances, changes in plant dynamics, and failure of components. The objective of this paper is to blend numeric-symbolic conversion techniques with linear conventional controllers so as to adapt to the environmental variations of the system. The control scheme is based on the parametrization of stabilizing controllers, which is called Kucera/Yula parametrization. The parametrization has been extended to the class of systems which contain numeric-symbolic converters. A neural network with a backpropagation training rule is used as the numeric-symbolic converter. It is shown how the numeric-symbolic converters can be blended with the linear controllers.