Nonlinear identification for diffusion/chemical vapor deposition furnaces

A simple nonlinear structure is proposed for system identification and uncertainty estimation of diffusion/chemical vapor deposition furnaces. The nonlinear model obtained using this structure approximates the behavior of furnaces in a larger operating region than the standard linear structures. The uncertainty estimates are used to quantify what kind of performance and robustness can be achieved by this model, and they could be used as controller design constraints or guidelines.; Another use of this nonlinear model is as a furnace simulator for educational purposes or for the development of high performance control strategies.; The proposed model structure and identification technique are applied on simulated data from simple first-principles models and experimental data from industrial furnaces. In both cases, the identified models are reasonable approximations of the actual systems. Moreover, the estimated uncertainty bounds indicate the feasibility of designing gain-scheduled controllers that meet the typical—for these processes—performance specifications.