ABSTRACT
This chapter is intended to provide a description of neural network control systems and of their potential for use in biomedical engineering control systems. Neural network techniques have been used by the engineering community in a variety of applications, with particular emphasis on solving pattern recognition and pattern classification problems [Grossberg, 1988a, b; Pao, 1988; Carpenter, 1989; Hecht-Nielsen, 1989; Sanchez-Sinencio and Lau, 1992; Zurada, 1992; Nerrand etal., 1993; Hagan et al., 1996]. More recently, there has been much research into the use of these techniques in control systems [Antsaklis, 1990; Miller et al., 1990b; White and Sofge, 1992]. Much of this research has been directed at utilizing neural network techniques to solve problems that have been inadequately solved by other control systems techniques. For example, neural networks have been used in adaptive control of nonlinear systems for which good models do not exist. The success of neural network techniques on this class of problems suggest that they may be particularly well suited for use in a wide variety of biomedical engineering applications. It should be emphasized that the field of neural network control is a relatively new area of research and that it is not intended to replace traditional engineering control. Rather, the focus has been on the integration of neural network techniques into control systems for use when traditional control systems alone are insufficient.
