Variability in the Lifetime of Maritime Thruster Components : An Exploratory Analysis of Component Breakdown

Abstract

In this thesis we conduct and exploratory analysis of the variability in the lifetime of maritime thruster components. The main objectives for the thesis have been to determine what causes the variability in the lifetime and to determine if weather variables can be one of the factors that affect the variability in lifetime. We have used a Random Forest classifier which is a method within machine learning. The Random Forest classifier was used to establish which characteristics that are important when determining if a component has sustained or failed. We also conducted a descriptive analysis to visualise and present the characteristics of the breakdown data. For the descriptive analysis have included data on seasonal weather patterns and locations for the breakdown of the components. This is to determine the relationship between them. We have concluded that the Random Forest classifier has a high probability of capturing whether a component sustain or fail based on the variables we have used to build it. The model reveal that a vessel’s physical characteristics are the best indicators of whether a component will fail or sustain its expected lifetime. From the descriptive analysis we have concluded that we do not have enough data or information to definitely establish the relationship between weather and the breakdowns of components. At the same time we don’t reject the possibility of a correlation between adverse weather and component failure.