ANALYSIS METHOD OF MAINTENANCE DATA IN ROLLING STOCK RELIABILITY ADMINISTRATION - RELIABILITY ESTIMATION FROM RANDOMLY LEFT TRUNCATED AND RIGHT CENSORED DATA IN RESTORATION-RENEWAL PROCESS.

To evaluate and reconstruct a rolling stock maintenance system, it is necessary to know the reliability of the rolling stock and/or its equipment. Statistical analysis of field maintenance data is a powerful tool for reliability estimation, but two methodological problems must be solved in order to perform such an analysis. One is the data incompleteness, due to observation during a limited period. The other is a failure-mixture situation where in case of failure the equipment is renewed or partially repaired to be used again. In this paper, a maximum likelihood estimation method is proposed for analyzing these kinds of data. Assuming a Weibull lifetime distribution, the estimation error of this method is evaluated by Monte Carlo simulation. A modified version of the Kaplan-Meier estimate and the hazard plotting method for the same data is also developed.