COMBINING LEAD FUNCTIONS AND LOGISTIC REGRESSION FOR PREDICTING FAILURES ON AN AIRCRAFT ENGINE

The increasing demand for performance and durability of advanced aerospace systems has increased the need for health management of these systems. Effective health management involves seamless integration of failure diagnostics, failure prediction, part life estimation, and maintenance logistics. These capabilities have only partially been implemented in current health management systems. Hence the effectiveness of current management systems has not achieved its potential. To achieve the goal of effective prognostic and health management (PHM), promising technologies from various disciplines must be integrated. One of these technologies is logistic regression. In this method, aircraft engine takeoff data is combined with control system fault information and by introducing lead times prior to the fault. Lead times of 1, 7, 14, and 30 days were analyzed using logistic regression on the difference from expected thermodynamic values. The resulting equations give probability of failure over time. An example using real engine data from GE-F414 engines to predict engine stall and anti-ice valve failures are presented. The results show good predictability of these events between a week and a month in advance. For example, for the event of an imminent anti-ice valve failure, the true-negative fraction was 99.6% and the positive-predictive value was 93.1%. This methodology can be combined with an engine health monitoring (EHM) system to provide prognostic failure predictions. Receiver Operator Characteristic (ROC) curves were evaluated as an additional measure of the quality of the predictions. These ROC curves show that there is prognostic value with this approach. This methodology can be updated and refined with additional data. As the results get more refined, the reliability of the fleet can increase, costs can be reduced, and safety increased.