Application and validity analysis of EM algorithm in reliability evaluation of aviation equipment

With the improvement of reliability of current aviation equipment, the field failure data is limited with few observations and random censoring observations. Aiming at the difficulty in reliability evaluation in this condition, a Monte Carlo simulation algorithm is employed to validate the feasibility of EM algorithm in the reliability evaluation of aviation equipments. The main thought is to assume a certain amount of aircrafts which can fly in a long time. The random censoring observations during the fault time are acquired with the dynamic change of the number of the aircraft, the flight years, and daily utilization rate. Together with Pauta criterion, the precision and validity of EM algorithm are analyzed quantitatively. A simulation example based on field data of a kind of aircraft shows that for this kind of aircraft, the validity of EM algorithm can reach 90% only if the sample size is above 20. Thus a larger sample size request is needed to improve the performance of EM algorithm.