基于机器视觉的煤矸石分选方法研究

    Research on coal gangue sorting method based on machine vision

    • 摘要: 传统的煤矸石分选方法存在成本高、效率低以及安全性不足等问题。近年来,随着深度学习相关技术的迅速发展,基于目标检测算法的智能选矸已经成为矸石分选的重要研究方向。为实现矸石与煤块的高效分选,本文提出一种基于机器视觉与深度学习相结合的检测方法。该方法以YOLOv5s模型为基础,首先,在主干部分中加入卷积注意力模块(CBAM)用于提高网络的特征提取能力;其次,在颈部网络部分采用加权双向特征金字塔结构(BiFPN)来增强网络的多尺度特征融合,避免漏检与误检现象的发生;再次,在预测部分使用EIoU函数作为改进后模型的损失函数,以进一步提高检测精度;最后,在训练前对原有数据集进行扩充,使模型的泛化能力得到进一步加强。实验结果表明:改进后模型平均检测精度为95.3%,较原模型提高了2.1%,能够有效地替代人工分选。

       

      Abstract: The traditional coal gangue sorting method has problems such as high cost, low efficiency, and insufficient safety. In recent years, with the rapid development of deep learning related technologies, intelligent gangue sorting based on object detection algorithms has become an important research direction for gangue sorting. To achieve efficient sorting of gangue and coal blocks, this paper proposes a detection method based on a combination of machine vision and deep learning. This method is based on the YOLOv5s model. Firstly, a convolutional attention module (CBAM) is added to the backbone to improve the network’s feature extraction ability; secondly, a weighted bidirectional feature pyramid structure (BiFPN) is adopted in the neck region to enhance the multi-scale feature fusion of the network and avoid the occurrence of missed and false detections; in the prediction section, the EIoU function is used as the loss function of the improved model to further improve detection accuracy; finally, before training, expand the original dataset to further enhance the model’s generalization ability. The experimental results show that the average detection accuracy of the improved model is 95.3%, which is 2.1% higher than the original model and can effectively replace manual sorting.

       

    /

    返回文章
    返回