论文部分内容阅读
Exponentially weighted moving average(EWMA)controller is recognized as the most popular run-to-run(RtR)controllers in semiconductor manufacturing process.As a model-based controller,the performance of EWMA-RtR controller depends on the quality of the process model and hence on the model-plant mismatch.There is an urgent need to evaluate the process model quality of EWMA-RtR control system.In this paper,a new performance index that is a minimum variance benchmark for the prediction error obtained from routine closed-loop data is proposed.Based on feedback invariant principle,the disturbance innovations are estimated by an orthogonal projection method.The proposed index is the ratio of a quadratic form of the estimated disturbance innovations and prediction error.Simulation example is given to verify the effectiveness of the proposed method.