Towards Distributed IoT/Cloud based Fault Detection and Maintenance in Industrial Automation

Abstract

Industrial Internet of Things (IIoT) automation should be based on a framework that guarantees flexible and energy efficient monitoring and control, without the need for frequent human intervention. The ability to analyse and process machine faults in real time is vital, however it poses many technical difficulties and challenges, mainly for industrial application environments. In our paper, we propose a novel, energy efficient, IoT and Cloud based decentralised framework for real time machine condition monitoring (MCM) and fault prediction, where computational demanding tasks are distributed across fog nodes and decision fusion rules are set and controlled by the Cloud. In particular, data acquisition phase is done by sensors distributed across machines, feature extraction and health condition classification is done by fog nodes, after receiving data and instructions as processed by the Cloud node. Our framework is based on collaboration and information flow among IoT, Fog and Cloud layers. To this purpose, we formulate a global consensus cross layer optimisation problem, concerning industrial healthy status monitoring, and we solve it in a distributed manner by applying asynchronous altering direction method of multipliers (ADMM) algorithm.