Big Data and Machine Learning for the Smart Factory—Solutions for Condition Monitoring, Diagnosis and Optimization

The increasing heterogeneity of automation systems coupled with the demand of more flexibility leads to a rising complexity of production plants. For this reason, production plants are becoming highly error-prone, and error detection is getting more difficult. Additionally, a great amount of expert knowledge is required to optimize production plants. Although machine learning approaches help to cope with these problems, these approaches are reaching their limits fast. As a result, approaches using Big Data are getting more popular, which help in finding relationships within the data, reliably detect anomalies (in advance), and allow an automatic optimization of the system. This contribution presents some approaches in the context of Big Data analyses in discrete continuous and hybrid production plants. This enables services in smart factories such as condition monitoring diagnosis and optimization. The contribution is rounded off with the presentation of practical examples of how the smart services can be applied.