Combining Bayesian belief networks with gas path analysis for test cell diagnostics and overhaul

Abstract

Engine overhaul shops need a reliable analytical methodology to pinpoint the cause(s) of engine test-cell under-performance to aid the overhaul decision-making process. Gas path analysis codes have been somewhat successful, but have not been entirely satisfactory. Previous works [Doel, 1994] have raised the idea that if other information could be integrated with the gas path analysis results, it may be possible to achieve better results This paper presents a diagnostic system developed for the CF6 family of engines. The system integrates test cell measurements and the gas path analysis program results with information regarding engine operational history, build-up workscope, and direct physical observations in a Bayesian belief network. The paper lays out the nature of the problem and the system requirements and design. The system produces a diagnosis while following a cost-effective diagnostic process using value of information calculations. This is illustrated through sample cases.