Automated Monitoring of Manufacturing Processes, Part 1: Monitoring Methods

Abstract

This paper presents a systematic study of various monitoring methods suitable for automated monitoring of manufacturing processes. In general, monitoring is composed of two phases: learning and classification. In the learning phase, the key issue is to establish the relationship between monitoring indices (selected signature features) and the process conditions. Based on this relationship and the current sensor signals, the process condition is then estimated in the classification phase. The monitoring methods discussed in this paper include pattern recognition, fuzzy systems, decision trees, expert systems and neural networks. A brief review of signal processing techniques commonly used in monitoring, such as statistical analysis, spectral analysis, system modeling, bi-spectral analysis and time-frequency distribution, is also included.