Coal Flow Detection of Belt Conveyor Based on FPN

View at publisher Summarise with VIRAT
Authors: Jinyang Zhang, Yilin Bei, Hao Wu
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

In coal mine production, real-time monitoring of coal flow on conveyor belts is of great significance for ensuring production safety and improving transportation efficiency. Existing coal flow detection methods primarily rely on parameters such as coal width, belt speed, and coal flow depth, employing volume modeling or mass estimation. These methods face challenges such as complex sensor deployment and high costs. Therefore, this paper proposes a coal flow detection method based on a Feature Pyramid Network (FPN), which models the dynamic changes in coal flow on the conveyor belt and determines the duration to achieve intelligent coal flow detection. The backbone network incorporates Partitioned Multi-head Self-Attention (PMSA) to enhance local modeling capabilities. The FPN structure includes Adaptive Fine-Grained Channel Attention (FCA) modules and Convolutional Block Attention Module (CBAM), effectively preventing information loss and enhancing responsiveness to critical spatiotemporal information. Experimental results demonstrate that this method achieves good detection performance across various scenarios, providing a decision-making basis for intelligent speed control of conveyor belts.

Fields of Study & Subjects

Belt Conveyor Systems Engineering Advanced Neural Network Applications Mining and Industrial Processes

Publication details

Journal / source
Journal of Vibration Testing and System Dynamics
Volume
10
Issue
3
Pages
283–296
Publication date
23 March 2026
Language
EN