New approches for structures detection and parameter estimation in non-linear systems identification
non-linear systems identification , NARX model, orthogonal least squares.
The identification of the process, during last decades, is becoming more important to the industry. Because of incrissing of the complexity of the loops and the needy of otimization of production. Althought, in most cases, mathematical linear models are used in industral sphere, some of it are not capable of adequately represent the dynamics of complexy systems. Thus, This paper objectives to present a contribution to non-linear systems identification area. In this thesis proposal are presented systems identification basic concepts, some methods based on Orthogonal Least Squares, as well as structures detection and parameter estimation of a NARX system by using these techiniques.