Knowledge-based System to Support Failure Cause Identification in Automated Manufacturing Lines

Abstract

The maintenance activities of automated manufacturing lines require a deep understanding of equipment and skills. One widely used failure analysis method is Failure Mode and Effect Analysis (FMEA). Previous research has attempted to utilize knowledge of existing failure phenomena based on past analyses to assist in identifying causes of failures. However, these studies have been limited to retrieving descriptions from knowledge bases and have not extended to searching for similar failure phenomena according to differences in manufacturing line processes. This study proposes a method to infer failure causes using embedding techniques by constructing a knowledge graph that structures the conceptual differences in manufacturing lines from FMEA descriptions.