Non-parametric analysis of maintenance data for Attitude Indicator of a commercial aircraft fleet

Abstract

Analysis of maintenance data for a repairable system provides information about the failure behavior of the system. Such information is needed for determining preventive maintenance and retirement policy for the system. Parametric and non-parametric models can be used for analysis. Parametric models require more assumptions about the failure process of the systems under consideration compared to non-parametric models. To verify these assumptions statistical expertise needed. The purpose of this paper is to show that in practice non-parametric estimator of mean cumulative function can be utilized easily to model the failure behavior of a fleet. Mean cumulative function estimates the mean number of failures as function of operating hours. The method is exemplified on the attitude indicator units of a commercial aircraft fleet. Sampling uncertainty of the estimates is quantified by normal approximation confidence intervals.