Cascade identification of nonlinear systems

Abstract

This work solves the problem of identification of nonlinear systems the standard implementation of which are the telecommunication systems. Nonlinearity of the reception and transmitting nodes of communication channels causes the intermodulation distortion of transmitted signals. Compensation of this distortion is based on the identification of nonlinear characteristics of the abovementioned nodes and their dynamic parameters. An analysis the traditional approach to the solution of a problem of nonlinear identification was made to obtain a solution based on the classical concept of “a black box”. It was shown that using this approach to produce the concentrated model of a nonlinear system with adaptation by its output signal faces inevitably the basic restrictions caused by incommutativity of nonlinear and linear nodes of a system. To overcome this restriction the we offer the stage-by-stage composition and identification of the nonlinear model parameters that are based on the principles of a multialternative structure and functioning of composite systems i.e.: multilevelness, modularity and separation of functions. The results of a pilot study of numerous variants of this model are shown here, viz., the concentrated model and multistage model with adaptation by output signal and a multistage model with adaptation by a vector of state. High efficiency of the model with stage-by-stage identification of nonlinear system was confirmed.