Predictive Maintenance using Incipient Fault Detection

Abstract

One focus of the Industry 4.0 paradigm is to enable Smart Factories with improved productivity and reduced down-times. In this context, Predictive Maintenance (PM) is a proactive approach to industrial services that optimises maintenance actions based on the system’s health. In order to monitor and understand the system’s status, effective PM requires dedicated tools capable of managing a large amount of data and discern the right data set required for analysis. As an aid for engineers, the software called MADe can be used. MADe is a model-based platform that can optimise maintenance actions following the information provided by the software itself, concerning sensor selection and functional models. In particular, among many others, MADe incorporates functionalities for incipient fault detection, which may be extremely useful when monitoring systems comprising fatigue or aging sensitive components. In fact, early fault detection enables scheduling of maintenance that will minimise the impact on production outputs. Owing to these considerations, this paper describes a technique for detection of incipient faults components affected by fatigue using an Equivalent Damage Index (EDI). This technique is tested on data taken from the literature in order to verify its potentials.