SYSTEM IDENTIFICATION BY CROSSCORRELATION METHOD USING PSEUDORANDOM SEQUENCES

Abstract

In this thesis a study is made of the cross-correlation techniques using pseudorandom sequences for the identification of linear, non-linear and multiinput/output time invariant systems. Limitations of using white noise as input test signal have been discussed . Two problems in using P.R.S. are investigated: (i) The non-zero off-peak value of the autocorrelation function of binary P.R.S. (ii) The effect of low frequency additive drift distu.,:e . A generalized method is developed to compensate for the non.zer000ff~peak value of the autocorrelation function of Unary P.R.S. It is shown that several of the existing methods are special cases of this generalized method. 'Three related methods are proposed to compensate for the effect of add-itive low frequency drift disturbances. In each of these methods it is assumed that the lour frequency drift can be described by a low-order polynomial. 1~io methods are proposed for the crosscorrelation identification of non-linear systems. In the first method each channel of the nonlinearity is identified separately by elimin• ating the others. Second method discusses with the orthogonal Li i) expansion of the output of the nonlinear system and then separately measuring *11 the lin.0 Kernels experimentally. The required properties of the input test signal for muitiinput/output system identification have been discussed. Practical applications of crosscorrelation identification using P.R.S. as input test signal are presented.