(ХНУРЭ, 2010) Бессонов, А. А.; Руденко, О. Г.; Руденко, C. О.
The given work describes some features of developing neural network models of technological processes. NARX and NARMAX neural network models are considered as an alternative to classical identification techniques. We describe a training procedure based on the theory of robust regression for dealing with outliers in the framework of function approximation, system identification and control. The procedure combines the numerical robustness of a particular class of non-quadratic estimators known as M-estimators in Statistics and dead-zone.