Prioritization of Responsive Maintenance Tasks via Machine Learning-based Inference

Maintenance task prioritization is essential for allocating resources. It is estimated that almost 1/3 of the maintenance cost is wasted to unnecessary activities. Task prioritization is based on risk assessment that takes into account the probability of failure and the criticality of asset (or consequence of failure). The criticality analysis is defined by the asset owner based on several parameters, among them safety, downtime cost, productivity, whilst the probability of failure is determined based on deterioration models, regular manual inspections, or sensors. The criticality of assets varies significantly between organizations, due to differences between their key performance indicators and maintenance objectives. Currently, the quantitative evaluation of the criticality of assets is a very complicated procedure for organisations. It depends on elaborate weighted score methods and extensive data collection efforts. However, the data required are not always available. This paper proposes an innovative method that exploits the advances in mobile communications, social networking, Internet of Things and machine learning to address this shortcoming. This approach brings building elements and assets online using asset tags with an online ‘asset profile’ linked to it. Users of assets are able to scan these tags using a mobile phone app to not only see the information about those assets, but also enter ‘comments’ describing issues and problems on the profiles. Natural language processing (NLP) is then applied to these c omments to estimate the criticality of assets. The proposed method is validated with historical data provided by the Estate Management, of the University of Cambridge.