Privacy-preserving condition-based forecasting using machine learning

As machines get smarter, massive amounts of condition-based data from distributed sources become available. This data can be used to enhance maintenance management in several ways, such as by improving maintenance demand forecasting and spare parts and capacity planning. Regarding the former, machine learning techniques promise substantial benefits for forecasting the demand for spare parts over conventional techniques that are commonly used. While development and implementation of these techniques is difficult, practical applications pose another important challenge to providers of maintenance, repair, and overhaul services. Their customers are reluctant to provide access to sensitive real-time data because of privacy concerns, and even more so when their data is stored and processed in the cloud. In this paper we describe an application for privacy-preserving forecasting of demand for spare parts based on distributed condition data. It combines machine learning techniques—more specifically, decision-tree classification—with order-preserving encryption. The application is appropriate whenever planning for spare parts for the maintenance of condition-monitored machinery is needed, and it is particularly suitable for cloud-based implementation. © 2018, Springer-Verlag GmbH Germany, part of Springer Nature.