Abstract:
The work included in this thesis deals with model reduction
tech ni.cl LIC S ill f' req tic ric,y ciom,iiil i.. c:. Ici:u'(I c,il :I I. Ii 1,1111c[ ic>ii description of the original system.
The first chapter introduces model reduction problem, its necessity and a broad classification of various model reduction techniques. This is followed ww a detailed procedure X minimization of a performance index, the Integral square cr. ror (iii:,) , in Chapter-? Stability based reduction methods are d/scribed in Chapter-3. The mixed methods to obtain reduced order model (ROM) are presented in Chapter-4, by combining the stability based reduction methods of Chapter-3 and error minimization technique. The denominator of the ROM is obtained by the stability based reduction, methods and optimal coefficients of numerator polynomial arc obtained by minimizing, the performance index, ISL;. The respective step-response of the illustrative examples ure Shown For comparison purposes. The mixed methods described in Chapter 1I , are extended to reduce the order of discrete-time systems in Chapter-5. A scheme to design a controller, using ROMs obtained from mixed methods, is given in Chapter-6.
The computer programs, in FORTRAN, for both continuous and discrete time case1;'a have also been devc l oped and =i.mplcmcrlted success-fully on a PC.
