Adaptive modelling of non-linear errors-in-variables systems

This paper discusses the adaptive modelling algorithm of the non-linear Errors-in-Variables (EIV) system. The non-linear EIV model is designed as a Wiener model with noisy input and noisy output measurements. The model consists of a moving average model linear subsystem and a piece-wise linear function as the non-linear subsystem. In order to ensure the online real-time updating of system parameters, the recursive form is designed for the algorithm to keep its adaptive properties and improve the parameters' identification efficiency. The derivation process of the algorithm ensures the recursive calculation results converge to their corresponding true values. The validity and convergence of the algorithm is confirmed by a simulation experiment with the help of Matlab.