A STATISTICAL STUDY OF SPECTROMETRIC OIL ANALYSIS DATA FROM THE NAVAL OIL ANALYSIS PROGRAM.

摘要

The thesis examines spectrometric oil analysis data from two sources in an attempt to formulate a statistical model which will be useful in monitoring aircraft engines in the Naval Oil Analysis Program. Initially,experimental data,gathered for an Air Force study,was used to determine if the measurement error inherent in the monitoring procedure is normally distributed and if correlations exist between measurements for different wear metals. Based on the results of this investigation, a study was made of operational data from Wright reciprocating engines of the R1820-82type. The investigation leads to the conclusion that a multivariate regression model is useful in estimating the parameters of the distribution of analyses from properly operating engines of this type. A procedure is suggested which would employ the readings from past oil analyses from a particular engine to determine its present condition. (Author)