Automating equipment productivity measurement using deep learning

Elham Mahamedi, Kay Rogage, Omar Doukari, Kassem Kassem
Northumbria University Newcastle, United Kingdom

DOI: 10.35490/EC3.2021.153
Abstract: Measuring the productivity of earth moving equipment help to identify their inefficiencies and improve their performance; however, measurement processes are time and resource intensive. Current literature has foccussed on automating equipment activity capture but still lack adequate approaches for measurement of equipment productivity rates. Our contribution is to present a methodology for automating equipment productivity measurement using kinematic and noise data collected through smartphone sensors from within equipment and deep learning algorithms for recognizing equipment states. The testing of the proposed method in a real world case study demonstrated very high accuracy of 99.78% in measuring productivity of an excavator.
Keywords: Excavators, Productivity, IMU, Machine Learning.
Pages: 140 - 147
Paper:
Contribution_153_final

Presentation Video: