DeepImpact: A Deep Learning Model for Whole Body Vibration Control using Impact Force Monitoring


Large capacity shovels are matched with large capacity dump trucks for gaining economic advantage in surface mining operations. These high impact shovel loading operations (HISLO) result in large dynamic impact force at truck bed surface. This impact force generates high-frequency shockwaves which expose the operator to whole body vibrations (WBVs). These WBVs cause serious injuries and fatalities to operators in mining operations. The present work focuses on developing solution technology for minimizing impact force on truck bed surface, which is the cause of these WBVs. The proposed technology involves modifying the truck bed structural design through the addition of synthetic rubber. Detailed experiments, with the technology implementation, showed a reduction of impact force by 22.60% and 23.83%, during the first and second shovel passes, respectively, which in turn reduced the WBV levels by 25.56% and 26.95% during the first and second shovel passes, respectively, at the operator’s seat. To make the smart implementation of the technology feasible, a novel state-of-the-art deep learning model, ‘DeepImpact,’ is designed and developed for impact force real-time monitoring during a HISLO operation. DeepImpact showed an exceptional performance, giving an R2, RMSE, and MAE values of 0.9948, 10.750, and 6.33, respectively, during the model validation. This smart and intelligent real-time monitoring system with design and process optimization would minimize the impact force on truck surface, which in turn would reduce the level of vibration on the operator, thus leading to a safer and healthier working environment at mining sites.


Mining Engineering

Keywords and Phrases

Artificial intelligence; Deep learning; Dump truck; Machine learning; Shovel dumping; Surface mining; Synthetic rubber; Whole body vibrations

International Standard Serial Number (ISSN)

0941-0643; 1433-3058

Document Type

Article - Journal

Document Version


File Type





© 2020 Springer-Verlag, All rights reserved.

Publication Date

27 Jul 2020