Abstract

In recent years, reliability, availability, maintainability and cost concepts have been widely applied as a tool for maintenance planning. This paper presents novel optimal preventive maintenance strategy for refuse collection vehicles (RCV), based on probabilistic and computational intelligence approach and suitable for companies with diverse vehicle fleets’ structure, making direct application of maintenance strategy recommended by the manufacturers of RCVs inappropriate. Three presented models for availability and cost optimization include a stochastic degradation process, random failures and a set of maintenance actions and their effects. First model analyzes strategy where maintenance is performed with a constant rate, independently from the system condition. In the second and third model maintenance rate has different value at each degraded state and it is dependent of degradation coefficient. The degradation/repair process is modeled as a time continuous Markov chain. Models were tested on actual maintenance and failure data for RCVs of large Serbian public utility company. The optimal value of the mean time to preventive maintenance at each degrading state was determined by using multi-objective genetic optimization. Presented models and estimation technique are useful for systems that degrade and cannot be restored back to “as good as new” condition with minimal maintenance activities.

Key words: Maintenance, refuse collection vehicles, computational intelligence, genetic algorithms.