Data-Driven Health Assessment in a Flight Control System under Uncertain Conditions

PHM technology plays an increasingly significant role in modern aviation condition-based maintenance. As an important part of prognostics and health management (PHM), a health assessment can effectively estimate the health status of a system and provide support for maintenance decision making. However, in actual conditions, various uncertain factors will amplify assessment errors and cause large fluctuations in assessment results. In this paper, uncertain factors are incorporated into flight control system health assessment modeling. First, four uncertain factors of health assessment characteristic parameters are quantified and described by the extended lambda-PDF method to acquire their probability distribution function. Secondly, a Monte Carlo simulation (MCS) is used to simulate a flight control system health assessment process with uncertain factors. Thirdly, the probability distribution of the output health index is solved by the maximum entropy principle. Finally, the proposed model was verified with actual flight data. The comparison between assessment results with and without uncertain factors shows that a health assessment conducted under uncertain conditions can reduce the impact of the uncertainty of outliers on the assessment results and make the assessment results more stable; therefore, the false alarm rate can be reduced.