A Process Mining Method Aware of Waist Event in Power System

The mining technology for power system process models is becoming increasingly mature. The aims of process model mining are to discover behavioral features in event logs and characterize them through process models. Compared to the traditional method of directly mining models from power system logs, focusing on waist events can fully leverage the advantages of special events in power system event logs. This utilizes the indispensable and potential sequential features of waist events in power systems, overcoming the limitations of traditional mining methods in analyzing log behavior features. A process model mining method is proposed that aware of waist events in the power system. The method first identifies waist events that have a special impact on model mining. Then, a feature set is designed so that the strict order relationship of modules can be reflected in the sequence of module sets. And, the feature set is improved by using event adjacency and behavior profiles, so that the new module set contains all elements in the entire event set and satisfies the module behavior profiles. Afterwards, expand the modules in the module sequence set according to the control-flow structure; Finally, the behavioral features of events in the process model are expressed through a process tree.