Industry 4.0 – Machine Learning from Big Data at Infineon
Industry 4.0 also known as Cyber-Physical Production Systems are systems where sensors and actuators are combined with communication and computation in order to achieve the optimal control of physical processes. Infineon – one of the world leaders in mixed signal chip production – has embarked in a very innovative and daring project the next steps for complete automation of one of its production sites. This automation involves the sensing, processing, and storing of huge amounts of data related to the processes involved in their chip production. The main goal of this project is to use big data and machine learning techniques to automatically construct a (stochastic) model of the entire chip production. This model is supposed to reflect the correlations between the measurements performed at the various stages. The model is supposed to be afterwards used for making predictions about the quality of the chip under construction and in case of anomalies to infer the interplay of the various processes responsible for these abnormalities. A further use of these models could be the automatic control of the machines involved such that the abnormalities are resolved on the fly. The master student is supposed to have a strong background in statistical techniques developed within machine learning, in particular, for big data, and to have solid background in programming. During the thesis the student should make a systematic survey of the research and practices developed in academia and in other chip production companies such as Intel, IBM, Honeywell, etc. The student is also supposed to spend some time at Infineon in Villach in order to get a deep understanding of the data and processes used by Infineon in their production. The interested student should contact Manuel Wimmer (firstname.lastname@example.org), Radu Grosu (email@example.com), Andre Kaestner (firstname.lastname@example.org).