Detection and Prediction of Faults in Industry using IoT Based SELF-TEL Architecture

Abstract

Usage of Fog networks in the Industrial Internet of Things (FIIOT) is the new upcoming
technology in industrial automation where machines communicate among each other for efficient data
transfer. The main important feature of FIIOT is predictive maintenance, which plays a prominent role and
also remains challenge in the field of industrial automation. Even though FIIOT plays an important role in
above applications, effective implementation with the best fit machine learning algorithms remains to be the
real challenge among the industries and researchers. Hence new intelligent architecture Self and Efficient
Learned Fogs (SELF) with Threshold based Extreme Learning (TEL) algorithms has been proposed for
predictive maintenance, which is considered to be best fit in Industry 4.0 which uses Fog Networks. This
paper aims at detection and classification of faults which can be carried out in the fog.The proposed
algorithm has been tested with Raspberry Pi 3 and Node-MCU and various parameters such as accuracy,
sensitivity, selectivity and fairness index has been evaluated. The results are more promising and it
outperforms other existing architectures