Integrated identification and control for diffusion/CVD furnaces

The paper describes an integrated approach for modeling and robust multivariable controller design, with application to temperature control of diffusion and CVD furnaces. The modeling relies on input-output data, collected in a preliminary identification experiment at the desired operating conditions. An important component of the identification is the computation of uncertainty bounds that describe the confidence limits of the model, in a manner consistent with robust control theory. The results of the identification are then used to design an H-infinity controller with a cascade, hierarchical structure. The final implementation of the controller also includes mechanisms to prevent integrator wind-up. Special emphasis is placed on automating the procedure -for classes of furnaces- so that, potentially, it can be used by non-experts in the field. Implementations include several industrial furnaces. Experimental results demonstrate the success of the approach as well as a good agreement between predicted and actual closed-loop behavior.