Abstract

Condition-based maintenance (CBM) is an advanced maintenance strategy in which maintenance actions are scheduled based on both the age data and condition monitoring information. Proportional Hazards Model (PHM) is a powerful statistical tool for estimating the equipment failure rate under condition monitoring. Effective CBM using PHM can decrease the overall maintenance costs by reducing unnecessary scheduled preventive maintenance actions. In CBM using PHM, main optimization objectives including minimizing maintenance costs and maximizing equipment reliability typically conflict to each other. But the reported research only focuses on single-objective. In this thesis, we propose a multiple-objective CBM optimization approach based on physical programming, which can systematically balance the tradeoff between the optimization objectives and find the optimal solution that best represents the decision maker's preference on the objectives. In CBM using PHM, the accuracy of parameter estimation greatly affects the accuracy of the model in representing and predicting the equipment health condition. Traditional optimization methods such as Newton's methods are inaccurate because they can only find local optimal value in parameter estimation. In this thesis, we develop an approach based on Genetic Algorithms (GA) for PHM parameter estimation and this approach can improve the accuracy of parameter estimation significantly. To illustrate the proposed approaches, we conduct two case studies using real-world vibration monitoring data, shearing pump bearings in a food processing plant and Gould pump bearings at Canadian Kraft Mill. The proposed approaches contribute to the general knowledge of condition based maintenance, and have the potential to greatly benefit various industries.