Intelligent Framework for Efficient Time Series Anomaly Detection in Industrial Applications Using Machine Learning
D J Chaithanya, P. Bharath, V Jaswanth, Sanjana Srinivasa, Vinayak Venkappa Pujeri, S. Shivani · 2025
Sensors are ubiquitous, but processing and generating insights from all that data is difficult. AI approaches have received significant attention in recent years, but most techniques have required significant computing resources, far beyond what could reasonably be incorporated into a small sensor. This paper builds on a successful work that demonstrated condition monitoring of a centrifugal pump using time series vibration data by broadening the created framework into a tool applicable to many more time series anomaly detection projects. This research work takes a deeper dive into the architecture of this framework and with the help of an alternative dataset, offers insights into how it could be leveraged to solve a variety of possible condition monitoring tasks that could be applicable for industry both for incorporation into future sensing products and with application to its vast array of manufacturing equipment.