摘要
传统重力分选煤矸石因精度低、依赖水资源等问题,难以满足高效分选需求。基于YOLOv8与混合注意力机制(CBAM)的煤矸石智能识别模型,在YOLOv8的基础上,通过引入CBAM模块,从通道和空间两个维度增强了特征提取能力。结果表明,模型在中等置信度范围(0.5~0.9)内实现了精确率与召回率的最佳平衡,F1分数达到峰值,平均精度(mAP@0.5)高达0.989。此外,模型在训练与验证阶段表现出良好的收敛性和泛化能力,验证集上的精确率和召回率均达到约0.98,进一步体现了其鲁棒性。为煤矸石的高效分选提供了理论依据和技术支持,有助于提升资源利用率。
Traditional gravity separation methods for coal gangue face challenges in meeting the requirements of efficient separation due to issues such as low accuracy and water resource dependency.This study proposes an intelligent coal gangue recognition model based on YOLOv8 and the convolutional block attention module(CBAM).By integrating the CBAM module into the YOLOv8 architecture,the model enhances feature extraction capabilities through dual-dimensional attention mechanisms in both channel and spatial domains.Experimental results demonstrate that the model achieves an optimal balance between precision and recall within the medium confidence range(0.5~0.9),reaching a peak F1-score with a mean average precision(mAP@0.5)of 0.989.Furthermore,the model exhibits excellent convergence and generalization capabilities during training and validation phases,maintaining precision and recall rates of approximately 0.98 on the validation set,thereby confirming its robust performance.This research provides theoretical foundations and technical support for efficient coal gangue separation,contributing to improved resource utilization efficiency.
作者
杨喜圆
张昕宇
Yang Xiyuan;Zhang Xinyu(Shenmu Zhangjiamao Mining Co.,Ltd.,Shaanxi Coal Group,Shaanxi,719316)
出处
《当代化工研究》
2025年第11期59-61,共3页
Modern Chemical Research
关键词
煤矸石
智能识别
YOLOv8
混合注意力机制
coal gangue
intelligent recognition
YOLOv8
convolutional block attention module
