ESTIMATING THE RELIABILITY OF CONDITION-BASED MAINTENANCE DATA USING CONTEXTUAL MACHINE-SPECIFIC CHARACTERISTICS

Abstract

In the mining industry, inter-connected machinery operates under harsh conditions 24 hours a day. Naturally, this degrades their state, and can lead to premature breakdowns and production losses. Condition-based maintenance (CBM) is a strategy that plans maintenance schedules depending on the condition of the equipment, and aims to improve decision-making processes. Data collected from machinery for CBM purposes must be reliable to avoid negative impacts on the maintenance strategy. Data reliability can be estimated by comparing multiple data streams; however, they are not always available, and can be expensive. This study aims to estimate the isolated and contextual reliability of single-source CBM data by applying multiple data analytics techniques. An application is designed to analyse current data on a machine level and to determine combined reliability. A case study implementation shows the difference in reliability classification accuracy between the isolated and contextual methods, highlighting the need for them to be combined.