Proportional hazards analysis of diesel engine failure data

摘要

AbstractCanadian Pacific Rail's diesel engines undergo scheduled major overhaul at intervals of approximately four years. Between these overhauls the condition of an engine is monitored by a series of physical tests and spectroanalysis carried out on samples of lubricating oil. At present the results of these tests are interpreted by the combination of skilled oil‐analyst technicians and an expert system. The success rate of this is over 70 per cent, where success is defined to mean that the oil‐analyst/expert system recommends that the engine should be scheduled for maintenance and it does indeed subsequently show evidence of an imminent failure. The purpose of the paper is to report on an examination of the method of proportional hazards modelling to determine whether or not PHM could improve on the accuracy of the oil‐analyst/expert system in determining the risk of failure of a diesel e