Premium
Recursive Bayesian state estimation method for run‐to‐run control in high‐mixed semiconductor manufacturing process
Author(s) -
Tan Fei,
Pan Tianhong,
Bian Jun,
Wang Haiyan,
Wang Weiran
Publication year - 2020
Publication title -
asian journal of control
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.769
H-Index - 53
eISSN - 1934-6093
pISSN - 1561-8625
DOI - 10.1002/asjc.1977
Subject(s) - context (archaeology) , matrix (chemical analysis) , computer science , inverse , semiconductor device fabrication , bayesian probability , state (computer science) , process (computing) , algorithm , mathematical optimization , control theory (sociology) , mathematics , engineering , control (management) , artificial intelligence , paleontology , materials science , geometry , wafer , electrical engineering , composite material , biology , operating system
One of the challenges in semiconductor manufacturing processes is the state estimation of a high‐mix production system. The traditional algorithm consists of constructing a context matrix based on the product fabricating thread. The state of the context matrix is estimated using the Moore‐Penrose pseudo‐inverse method. Although the method works well, the context matrix is often singular. Taking an integrated moving average disturbance into consideration, a novel state estimation method is proposed in a high‐mix manufacturing scenario. Furthermore, the recursive Bayesian estimation is presented to obtain the estimations of states combined with a moving window and an analysis of variance model. As a result, the calculation of the inverse of the context matrix is avoided and the unobservability problem is addressed. Both simulated and industrial cases are presented to demonstrate the effectiveness of the proposed algorithm.