Feature Estimation for Punching Tool Wear at the Edge
Watch a 17-minute conference talk from the Eclipse Foundation's eSAAM 2023 event where Dr. Olli Saarela, Principal Investigator and Senior Scientist at VTT Norway, explores sheet metal punching tool wear monitoring through edge processing techniques. Learn about the comparison between two open-source time series feature extraction methods - TSFEL and MiniRocket - for building classification models based on acceleration measurement data. Discover how MiniRocket algorithm achieves superior classification accuracies of up to 96.5% compared to TSFEL's 35-56% range, making it a promising solution for real-time punch tool monitoring in automotive, aerospace, electronics, and construction industries. Understand the importance of effective tool wear monitoring for maintaining product quality, reducing scrap, and optimizing manufacturing efficiency, while acknowledging the need for further research with larger datasets.