Surveying on Big Data and Predictive Analytics – Based Machine Learning for Smart Industrial IoT Applications

Abstract

The progress of Industry 4.0 has allowed building intelligent factories with the application of artificial intelligence (AI), big data, and the industrial internet of things (IIoT) to boost system intelligence, automation, and efficiency. The IIoT devices produce very large amounts of heterogeneous and high-velocity data, and scalable big data architectures and sophisticated analytics are necessary to find actionable insights. Machine learning (ML) models, including 
Random Forest, Support Vector Machines, and Long Short-Term Memory networks, are crucial for process optimization, problem detection, quality control, and predictive maintenance. Nevertheless, the combination of big data and IIoT brings a number of problems such as data heterogeneity, real time processing limitations, model interpretability, security and privacy issues. This article highlights the uses and difficulties of big data-driven IIoT applications, reviews the lifecycle of big data in IIoT contexts, and discusses some of the most commonly used ML-based predictive analytics. The results  highlight the potential of introducing change to IIoT in relation to ML, but also the necessity of implementing sophisticated data management,explainable 
AI, and secure architectures to achieve the potential of smart industrial environments.