Development of a data-driven dynamic model for a plasma etching reactor

Abstract

We compare several parameter identification methods for data-driven identification and validation of an empirical linear dynamic model for a helicon plasma reactor. The model relates easily measurable process variables to ellipsometry measurements from which the etch depth can be determined in real time. The potential use of such a model for process control is obvious. The model developed shows improvement over a neural network model developed in a previous study based upon the same data.

Efficient nonlinear modeling of rainfall-runoff process using wavelet compression

“…In this paper, the wavelet used for analysis herein is the quadratic spline (QS) wavelet, which belongs to the BattleLemarie group (Zhang et al, 2002). The QS-wavelet has only one vanishing moment (Ge and Mirchandani, 2003).…”

“…To obtain good quality of compression and recovery during the procedure of signal compression, orthogonal wavelet base is usually being substituted for biorthogonal wavelet base (Wang and Yang, 2004). The main reasons for our choice are the biorthogonality of the QS-wavelet and its dual, and good interpolation properties (Zhang et al, 2002).…”