Event Pattern Constructing Technique for Prediction of Violations in Technological Processes

Abstract

The problem of the prediction of violations in business processes is considered in the paper. The authors suggest application of the frequent item sets mining technique for analysis of the event streams. This approach allows investigation of the event sequences and revealing of the event sequences which lead to the violations. The detected violation patterns then could be used for monitoring data in the real time mode. The proposed technique could be applied when analyzing data from sensors of different type and used in distributed systems. The method is able to predict a violation before it occurs avoiding thus dangerous situations.