Segmentation of Multi-State Compound Waveform and Extraction of Features for Anomaly Detection

In a semiconductor fabrication plant, various types of sensors are installed at various equipment of various processes to monitor the quality of the products and the progress of various processes. These sensors generate a huge volume of time-series waveform data, which are used for failure prognostics, failure diagnosis, and anomaly detection using supervised or unsupervised classifiers. For this purpose, automatic extraction of features from a huge volume of waveform data is needed; however, extraction of effective features automatically from a compound waveform having multiple spikes, complex states or transitions is difficult. If a compound waveform is properly segmented into multiple sub-waveforms, effective features from sub-waveforms can be extracted. In this paper, we propose a new waveform segmentation method based on two-step state and change-point detection using Kernel Density Estimation (KDE) and clustering techniques and apply IEEE Standard-based methods with our state determination technique to extract features from sub-waveforms. We apply our proposed method to waveform data of real sensors installed at one of our semiconductor fabrication plants and compare the performance with the conventional step-based segmentation technique and another pattern matching technique. The experimental results show that our proposed method enables extraction of effective features, which result in higher accuracy of anomaly detection compared to conventional techniques.

I. Introduction

In a manufacturing plant, power plant or factory, various types of sensors are installed at various equipment of various processes to monitor the quality of the products, the progress of various processes, the condition of the output, etc. These sensors generate a huge volume of time-series waveform data. There are a number of characteristics of this huge volume of time-series waveform data. First, the time-series waveform data are of variable lengths. Second, a waveform may transit among multiple stationary states depending on various processes but on a stationary state the waveform may have fluctuations, and during these transitions, it may produce spikes, overshoots, and undershoots. Third, there may be a minute difference between an abnormal waveform and a normal waveform. Fourth, the number of abnormal waveforms may be very small compared to the huge volume of normal waveforms because in fabrication of semiconductor and storage products, anomaly rarely occurs in an equipment and if an anomaly occurs, counter-measures are enforced to prevent re-occurring of the same anomaly. An anomaly detection technique that can detect anomaly using this huge volume of time-series waveform data at high speed and high precision is needed.