Description Method and Failure Ontology for Utilizing Maintenance Logs with FMEA in Failure Cause Inference of Manufacturing Systems

In the realm of manufacturing systems, inferences of failure causes have been performed mainly depending on the reuse of Failure Mode and Effect Analysis (FMEA). However, achieving inference with the same level of quality as experts has been challenging. The objective of this study is to improve the quality of inference by combining maintenance logs and FMEA in the inference of failure causes of manufacturing systems. There are two challenges in using maintenance logs for inference. First, it is difficult to extract causal relationships among failures because the format and quantity of maintenance logs are not consistent. Second, the hierarchy of failures described in each item of maintenance logs is not fixed, making it difficult to search for similar failures by using the items. To address these challenges, we propose a description method for maintenance logs that describes causal relationships among failures and those between failures and functions. We also introduce a failure ontology that represents the hierarchy of failures and conditions impaired by failures based on expert knowledge of manufacturing systems. For the two assumed failures, the inferences derived from maintenance logs and FMEA using the proposed method show better quality than the inferences using FMEA alone. The recall calculated from the failure cause candidates enumerated by experts increased by 3.7 and 5.0 times, respectively.