Logo image
Using wavelet-based method and multiway PCA for fault detection and fault diagnose via an empirical study in semiconductor manufacturing
Conference paper

Using wavelet-based method and multiway PCA for fault detection and fault diagnose via an empirical study in semiconductor manufacturing

Chia-Yu Hsu and Chen-Fu Chien
36th International Conference on Computers and Industrial Engineering, ICC and IE 2006, pp.258-267
2006

Abstract

Fault detection Fault diagnose Multiway PCA Semiconductor manufacturing Wavelet
The approach is proposed to multivariate statistical process control (MSPC) for process monitoring to improve the performance of existing model in semiconductor manufacturing. This study aims to propose fault detection and diagnosis approach in MSPC by applying wavelet transformation and multiway principal component analysis (MPCA) models. Wavelet transformation is used for denoising of the original data to extract deterministic pattern and approximately decorrelate measured data. MPCA model can be captured the few variables with the most explained variation and extract cross-correlation between variables. For process monitoring, both D statistic and Q statistic are as indicators of fault detection and contribution of each variable can direct the deviation of fault. To validate the proposed approach, an empirical study is conducted and the results showed practical viability of this approach.

Metrics

1 Record Views

Details

Logo image