Anomaly Detection in Workstation Using Deep Learning Techniques

EasyChair Preprint 11113, version 2

Abstract

Industry 4.0 has led to the development of smart manufacturing with control systems for data collection, optimization, and fault detection and diagnosis (FDD). However, building and setup, with regards to assembly lines, controls etc, contribute to significant global energy consumption. Digital Twin (DT) technology offers a sustainable solution for facility management and predictive maintenance of machinery. For this, Data-driven methods are gaining popularity due to their ability to handle large amounts of data and improve accuracy, flexibility, and adaptability. Also, Deep learning methods can analyze large and complex datasets, making them a promising area for further investigation in anomaly detection and other fields of Industry 4.0. This paper will be focussing on anomaly detection in the Workstation-1 present at the IAFSM Lab at IIT, Delhi. The workstation is integrated with many prime fields of Industry 4.0 like IIOT. Though this helps in increasing productivity, Data Collection, reducing operating costs as well as helping with predictive maintenance, at the same time makes it susceptible to an external attack over the cloud, in other words, makes it vulnerable to cyber-attacks. This can be proved very detrimental to the whole workstation, as an external attack (hacking) can influence various factors of the total operation, like changing the direction of the assembly lines, changing the inventory(using the incorrect raw material), or increasing/decreasing the pace of the whole process. All this can increase the overall operational cost and manipulation of the product’s quality. To counter this, this paper introduces a combined approach of Digital Twin and Deep learning to detect anomalies in the movements of the operating arm, as well as inventory control.

Keyphrases: Digital Twin, Industry 4.0, Smart Manufacturing, deep learning

@booklet{EasyChair:11113,
  author    = {Akshit Shishodia},
  title     = {Anomaly Detection in Workstation Using Deep Learning Techniques},
  howpublished = {EasyChair Preprint 11113},
  year      = {EasyChair, 2023}}