Development of Outlier Detection Algorithms for Sensors with Time-Varying Characteristics

In order to maximize production in advanced manufacturing industries, such as semiconductor manufacturing processes, equipment must perform consistently across multiple process steps. To ensure this, a process called tool-to-tool matching (TTTM) is used to align the performance of different equipment by analyzing sensor data. Although various TTTM techniques exist, there is no universally effective method. To address this, this paper aims to develop algorithms for detecting outliers from sensors that change over time by considering the unique characteristics of each sensor.