STABILITY CONSIDERATIONS OF REDUCED ORDER MODELS OBTAINED BY PADE~TYPE APPROXIMATIONS

Abstract

The low order modelling techniques in frequency domain for reduction of large order systems are reviseod. Drawback of the Moment matching technique, end a or tinued traction expansion and truncation method is pointed out. The draw back associated with the above mentioned methods is that the reduced model (models) may not sometimes be stable. Yn those oases where the methods produoi unstable reduced—order els, modifications in the methods of obtaining reduced order models suggested by several authors are pointed out. Further a testing procedure for the stability of low-order reduced models hex been proposed with the knovw-ledge of coefficients of expansion of the given system transfer function about SO. This helps in rejecting the unstable reduced order models. In addition to thin, a met of selection rules for stable form of reduced model a for given transfer function are obtained. These zu1es ale framed by taking large number of different form of system transfer functions. and their reduced models of second order., These rules describe how the stability of reduced models to effected by the coefficients of the numerator polynomial of the given sy step transfer f'uno tion.. Thus it assist* In selecting the stable form of second order model if exists.. However analytical proof for the selection rules is quit* difficult._