ABSTRACT
A ctive control of large-scale, flexible, reconfigurable structu res is problem atic. W hile a conventional approach m ight develop a control stra tegy from the governing equations of th e p lan t, there are problem s associated w ith th is m ethod. A neural network based control paradigm , inspired by biological system s, provides an alternative. This paper characterizes the classical neural com putational m odel in term s of the control problem and describes prior work in developing neural applications for control, as well as for re la ted problem s of in terest w ith in the structu res dom ain. The paper addresses the issues of applicability and use of neural com puting for s tru c tu ra l control and m onitor ing, and considers th e im pedim ents to building such a system . Conclusions argue th a t a neura l com puting approach to s tru c tu ra l com putations m erits fu rther exploration.
