From fault trees to diagnostics knowledge base: case study and quantitative improvement assessment

摘要

The level of readiness and availability of equipment and systems is strongly affected by troubleshooting and problem resolution quality. Until recently, the most pervasive method of supporting technicians in achieving high problem resolution performance has been the creation of fault trees. It is now accepted that knowledge-based problem resolution has many advantages over the fault tree method, including a vast acceleration in creating the troubleshooting support tool, easier modification, higher accuracy, and learning capability. However, there still exists a large body of problem resolution knowledge "frozen" into fault charts. This paper describes a systematic and automatic method of converting this knowledge into a model-based diagnostics knowledge base and describes one case in which the ClickFix intelligent troubleshooting support software was used to facilitate this conversion. The results prove the validity of this method and show the advantages in terms of automatic minimization of time required to complete fault isolation as well as increased diagnostics accuracy and power to handle unforeseen cases and multiple faults, and in terms of ease of modification. The results of this empirical study are compared to theoretical studies of similar methods.