Artificial intelligence of things based approach for anomaly detection in rotating machines

• Published in: Computers & electrical engineering Vol. 109; p. 108760
• Main Authors: Mian, Tauheed, Choudhary, Anurag, Fatima, S., Panigrahi, B.K.
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.07.2023
• Subjects: Edge computing; Fault detection; Internet of things; Machine learning; Remote monitoring; Internet of things; Fault detection; Remote monitoring; Machine learning; Edge computing
• ISSN: 0045-7906; 1879-0755
• DOI: 10.1016/j.compeleceng.2023.108760

•An artificial intelligence of things (AIoT) based approach is implemented for anomaly detection in rotating machines.•Edge centric approach is applied using a machine learning-based testing framework.•Real-time anomaly detection and control of rotating machines from a remote place. In the present era of Industry 4.0, Artificial Intelligence (AI) and Internet of Things (IoT) are revitalizing the predictive maintenance systems for industrial rotating machines through real-time approaches. In this work, an Artificial Intelligence of Things (AIoT) based framework is developed and implemented for anomaly detection in rotating machines through vibration monitoring. Signals are captured using a low-cost MPU6050 accelerometer with Raspberry Pi 4B as an edge device. The edge-centric mechanism is based on Support Vector Machine (SVM) for effective implementation on the edge device. A dedicated web-based platform is designed to communicate with the edge device for real-time decisions. The developed framework is validated on three rotating machines, including gear test rig, bearing test rig, and induction motor. The results demonstrate the robustness of the proposed method for anomaly detection on different machines. Also, the developed framework has the capability to remotely monitor and control the machines in real-time. [Display omitted]