Anomaly Detection Using Model Generation for Event-Based Systems Without a Preexisting Formal Model

Detecting and debugging faults more efficiently can significantly improve the performance of systems, and a first step toward fault detection is anomaly detection. A new anomaly detection solution is proposed in this paper for event-based systems that consist of processes that interact through shared resources and that do not have a preexisting formal discrete event system model. This solution generates models of the system, assesses the models' performance in detecting faults, and then uses the models and their performance to detect anomalies in new event streams. A new resource-based Petri net formalism is introduced to model these types of systems. The model generation uses an algorithm based on workflow mining to generate resource-based models. The proposed solution is demonstrated on two manufacturing cell examples.