Block-Oriented Identification using the Best Linear Approximation: Benefits and Drawbacks

Due to their simplicity and structured nature, block-oriented models are popular in nonlinear modeling applications. A wide range of block-oriented identification algorithms were developed over the years. One class of these approaches uses the so-called best linear approximation to initialize the identification algorithm. The best linear approximation framework allows the user to extract important information about the system, it guides the user in selecting good candidate model structures and orders, and it proves to be a good starting point for nonlinear system identification algorithms. This paper gives an overview of the benefits and drawbacks of using identification algorithms based on the best linear approximation.