Machine learning with SHapley additive exPlanations for evaluating mine truck productivity under real-site weather conditions at varying temporal resolutions

ABSTRACT

This study built truck productivity prediction models incorporating real-site weather conditions at varying temporal resolutions. The best models were combined with SHapley Additive exPlanations to offer quantitative and qualitative analysis for input variables’ effect on the model outputs. The results showed that mining engineers can make more accurate predictions of truck productivity at the weekly resolution compared with other resolutions. The three most influential input parameters were haul distance, empty speed, and ambient temperature. Extreme weather, such as strong wind speed, heavy precipitation, and extreme relative humidity, had a certain effect on truck-shovel allocation. Meanwhile, a unified graphical user interface was developed to predict mine truck productivity.

KEYWORDS:

• Mine truck productivity
• extreme weather conditions
• temporal resolutions
• machine learning
• SHAP analysis
• graphical user interface

Acknowledgments

The authors are grateful to Dr. Rui Qin at the University of Alberta for his helpful guidance on the graphical user interface.

Disclosure statement

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Author contributions

Chengkai Fan: Conceptualisation, Methodology, Software, Writing – Original draft, Writing – Review & Editing, Visualisation. Chathuranga Balasooriya Arachchilage: Conceptualisation, Methodology, Writing – Review & Editing. Na Zhang: Conceptualisation, Methodology, Writing – Review & Editing. Bei Jiang: Conceptualisation, Methodology, Resources, Supervision, Writing – Review & Editing. Wei Victor Liu: Conceptualisation, Resources, Supervision, Writing – Review & Editing, Funding acquisition.

Additional information

Funding

This work was supported by the Collaborative Research Project [RES0043251] and the Pilot Seed Grant [RES0049944] from the University of Alberta.