Performance evaluation of nonlinear automated model generation approaches for high level fault modeling

It is known that automated model generation (AMG) techniques are sufficiently mature to handle linear systems. Other AMG techniques have been working reasonably well for various levels of nonlinear behavior. However, most of the modeling are performed under MATLAB environment. To be more realistic, the models need to be translated into hardware description language (HDL) models, such as VHDL-AMS or Verilog-AMS models, to perform high level modeling (HLM) and high level fault modeling (HLFM), which is a challenging task due to its nonlinear behavior. In this paper, the capability of System Identification (SI) based nonlinear AMG techniques is investigated by converting MATLAB models into VHDL-AMS models and to perform HLFM. Several faults are modeled successfully in MATLAB environment using AMG. However, they failed to perform HLFM when run in HDL simulator SystemVision.