Anomaly Detection and Analysis of FDC Data

With the progress of semiconductor manufacturing technology and the continuous development of software and hardware technology, the sampling rate of the FDC (Fault Detection and Classification) system can be up to 100Hz. For automation Fab, higher sampling rate means higher detection rate of process anomaly detection. However, the high sampling rate has been the bottleneck and difficulty of current data processing techniques. PCA (Principal Component Analysis) is an algorithm which can discard the components caused by noise and be applied to feature extraction. This article mainly studied how to use PCA to reduce the dimension of FDC data for the anomaly detection, and establish a reasonable parameter eigenvalue database for converting FDC data to Euclidean distance which can be applied to various mainstream data analysis algorithm, for the aim of fast analysis of FDC data.