Identification of Human Operation Deviation Using a Conformance Checking Technique

Abstract

Human errors and procedure non-compliance are the main contributors to accidents and incidents in nuclear industries. In this paper, a token-based conformance checking method is applied to the underlying human error identification based on the process mining and audit analysis from a retrospective perspective. First, the sequences of human behaviors involved in task implementation are collected from operation log files. Then the nominal task process is modeled using Petri net in accordance with the standard operating procedures. Finally, a token-based conformance checking method is adopted to interpret the operation logs and identify the non-compliance behaviors such as error of omission or error of commission, etc., that may occur during task execution. The method is illustrated with an example case study of manual makeup of boron acid during a refueling outage in nuclear power plant. The case scenarios are simulated in a Defense-in-Depth risk monitor. The case analysis results show that the proposed conformance checking technique is effective for human error investigation.