Study of abnormal events relevance in process industry based on sequences pattern mining

By analyzing the relevance and coupling of abnormal events in process industry, this paper studies the feasibility, research background and important parameters of abnormal events sequences pattern mining based on chemical plants' DCS alarm data. According to the temporal relevance of the abnormal events, by adding time properties of abnormal events and improving the efficiency of the algorithm, we design and implement TFPG algorithm based on FP-Growth, which fetches temporal relevance rules from alarm data. Finally, the techniques we proposed are validated by analyzing the alarm data of accidents of a coal chemical group.