An overlapping receding horizon approach to reduce delay of disturbance detection and classification using Bayesian statistics

As the semiconductor industry is moving toward more flexible manufacturing processes and the device dimensions decrease, control strategies and controller algorithms for flexible manufacturing processes are needed to maximize process capability and quickly recover processes after process changes and disturbances. Currently EWMA is the most widely applied run-to-run controller due to its simplicity and robustness. However, because the same weighting is applied to the new measurement no matter whether it is a normal measurement or an outlier, the controller would track step changes slowly if the controller is tuned to reject noise well. In this work, a novel approach is developed based on Bayes theorem to address this problem. By apply an overlapping receding horizon approach, the developed algorithm can detect and classify the disturbance without additional delay. The performance of the proposed algorithm is demonstrated using both simulation and industrial examples.