Feature Extraction from Equipment Sensor Signals with Time Series Clustering and Its Application to Defect Prediction

Abstract

In semiconductor manufacturing processes, it is important to quickly identify any signs of the occurrence of defects. We applied a time-series clustering method to the signal data of processing equipment and obtained information related to the occurrence of defects. By using the information as the feature values of a prediction model, we were able to predict defects more accurately than by using only conventional feature values.