AI Enabled Manufacturing: A Deep Learning Approach to Network Fault Detection

Abstract

Fault-tolerant control of industrial robotics network and highly connected machines setup for manufacturing in a production line is crucial. In this paper, we will discuss the recent advancements in fault-tolerant control strategies for collaborative robots with focus on the communication network faults. Cyber-attacks on robotic cyber physical systems (CPS) in the context of fourth industrial revolution constitutes a threat to the modern production systems which requires real-time detection so that the damage to the physical layer could be avoided. By selecting appropriate features for the deep neural network (DNN), it has been found that, an accuracy of 94.64% can be achieved for classifying malicious attacks. Thus, artificial intelligence (AI) can play a substantial role in securing future industrial manufacturing systems from cyber-threats thus avoiding down time in the production lines and large scale manufacturing operations.