Initial Estimation of Wiener-Hammerstein System with Random Forest

Abstract

The most popular class of Volterra nonlinear dynamical system is Wiener-Hammerstein system. A static nonlinearity, positioned between two dynamical sub-system constructs Wiener-Hammerstein (W-H) system. The best linear approximation (BLA) technique assembles two linear filters and the nonlinearity into a single filter for input and noisy output. The identification challenge resides in separating two filters. This work proposes Random Forest (RF) as a first alternative to do this. It is like the selection of holiday destinations based on the recommendation of random traveller. The proposed technique supports reasonably high noise level and does not require to optimize all models. Thus, a speedup in processing time is achieved without any prior knowledge about the model configuration.