Predictive Modeling of Ball Mill Load Parameters Using Hybrid Physics-Informed Neural Networks

Conference Paper (2025)
Author(s)

Yiwen Liu (Taiyuan University of Technology)

Gaowei Yan (Taiyuan University of Technology)

Yusong Pang (TU Delft - Transport Engineering and Logistics)

DOI related publication
https://doi.org/10.1109/CCDC65474.2025.11090212 Final published version
More Info
expand_more
Publication Year
2025
Language
English
Pages (from-to)
1604-1609
Publisher
IEEE
ISBN (electronic)
979-8-3315-1056-5
Event
Downloads counter
23
Reuse Rights

Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.

Abstract

Wet ball mill plays a key role in the grinding process, and its load state directly affects production efficiency, energy consumption and product quality. Aiming at the problems of poor interpretability of pure data-driven models and complex modeling of mechanism models under variable working conditions, a hybrid prediction model DAPINN combining deep learning and physical information is proposed. By introducing the deep hidden physics model principle and using the characteristics of neural networks to approximate arbitrary functions to simulate complex physical partial differential equations, the physical interpretability of the model is enhanced. At the same time, the model introduces domain adaptation technology to improve the prediction accuracy and generalization of the model under variable working conditions. Experiments were conducted on data collected from a small ball mill in the laboratory. The experimental results show that under variable working conditions, the prediction accuracy of the DAPINN model is better than that of the pure data-driven model.

Files

Predictive_Modeling_of_Ball_Mi... (pdf)
(pdf | 0.85 Mb)
- Embargo expired in 05-02-2026
Taverne