An IIoT approach for edge intelligence in production environments using machine learning and knowledge graphs

Abstract

The mining industries need novel solutions to reduce production stoppages. Predictive maintenance solutions and especially the hardware components, cannot operate properly under such harsh conditions, as high concentration of dust and other chemical material may lead into fault sensor measurements. This study presents a solution for condition monitoring and predictive maintenance and productions status monitoring for assets operating in harsh operating environments. First, an edge device is connected to multiple sensors monitoring critical parameters related to the operating conditions of an asset. In particular, the device is in charge of data collection, filtering and smart data generation for further analysis and processing. At a later stage, the collected data are pushed to a cloud platform where predictive analytics, as well as production status analytics, are estimated. The condition monitoring and predictive maintenance component utilizes machine learning in order to estimate the Remaining Useful Life of the monitored asset(s). The production status component utilizes knowledge graphs that are populated with data provided by the edge device. The combination of these two components aims to provide meaningful insight to field personnel supporting them in decision making and production supervision. A prototype IIoT system has been implemented and tested in a use case related to an aluminium producing company with its results demonstrating the applicability of the proposed solutions for real-world application.