PHM, Prognostic and Health Management Software in Semiconductors

Organization: Industrial Technology Research Institute (ITRI)
Year: 2017

PHM, Prognostic and Health Management Software in Semiconductors uses AI-based machine learning algorithms to create the world’s first successful anomaly prognosis system for application in semiconductors. In today’s semiconductor industry, if an anomaly occurs during operation, the experience of equipment engineers is relied on to find the source of the problem, which can take four to five days and influence production capacity and quality. The concept and technology for semiconductor anomaly prognosis has been developing for over 20 years, but remains rare because prediction accuracy has not reached requirements for manufacturer use and application extension is difficult. PHM analyzes the ideal maintenance time via a machine learning approach, thereby reducing the frequency of unexpected shut down and repair, allowing factory operations to become more efficient. The software can also collect and analyze factory data, providing early warning notices before a problematic incident occurred. Early warning allows manufacturing engineers to adopt corrective measures before damage is done, which is beneficial for maximizing production and for ensuring the product quality.