Violation Detection in Heterogeneous Events Streams

Abstract Violation detection at production facilities is a vital task, especially for critical infrastructures. Nowadays, many monitoring sensors of different types are used for the event collection. These sensors often provide a data in different format making thus complex analysis of such data more difficult. This paper suggests an approach for such data analysis, its distinctive feature is grouping events and application of an individual classification method to each group. The suggested approach is demonstrated by its application on real data within continuous time period.