Improving test strategies and fault isolation with expert systems

This paper focuses on the use of commercial off-the-shelf (COTS) expert systems in integrated diagnostics (1D) for military applications. Expert systems have developed and matured over the past several years to become a viable tool capable of functioning as a procedural tool for identifying diagnostic requirements, analyzing test system capabilities, and providing seamless diagnostic data transfer from requirement to analysis to operations. The most important differentiating characteristics of expert systems are their modeling methods, and their architecture. The modeling method drastically affects the time required to build a model, and the architecture must be open enough to integrate with the many tools used in engineering, deployment, and maintenance of the supported equipment throughout its life cycle. In this article, we present the Fault Modeling method, which has been field-proven over the past decade as flexible enough to meet the challenges of different lifecycle tasks, as well as lending itself to learning-self-improvement over time, even when starting with no knowledge. Expert systems using this model feature rapid deployment, and are able to cover the entire ID process including: capture of existing data, analysis of fault detection and isolation capabilities of the unit under test, and a means to assess diagnostic system designs early in the development phase. The systems integrate easily with simulators, automatic test equipment (ATE), and portable maintenance aid (PMA) equipment.