摘要
随着世界工业化的发展,自卸卡车可极大地降低人力成本,加快工作效率,在露天矿区得到了广泛应用。然而,卡车电动轮的主机架在使用约2万~4万h后易出现开裂现象,需要进行维修。目前的裂纹检测主要通过人工目测和物理方法进行。然而,这些方法在日常维护时难以实现,无法满足快速、实时检测的要求。针对现有算法的问题,开发了基于改进YOLOv9的自卸卡车电动轮主机架裂纹检测算法。对收集到的图像进行了人工标注,并制作成数据集。模型使用YOLOv9网络并将骨干网络中Conv替换为Ghost Module,减少模型参数与冗余计算。实验结果表明,该模型识别准确度为93%,在保证实时性的基础上,具备较高的准确度。
With the development of world industrialization,dump trucks can greatly reduce labor costs and accelerate work efficiency,thus being widely used in open-pit mining areas.However,the main frame of the truck's electric wheels is prone to cracking after approximately 20000 to 40000 hours of use and requires repair.At present,crack detection is mainly carried out through manual visual inspection and physical methods.However,these methods are difficult to implement in daily maintenance and cannot meet the requirements of fast and real-time detection.We have developed a crack detection algorithm for the main frame of electric wheels in dump trucks based on an improved YOLOv9 algorithm to address the issues with existing algorithms.We manually annotated the collected images and created a dataset.The model uses YOLOv9 network and replaces Conv in the backbone network with Ghost Module to reduce model parameters and redundant calculations.The experimental results show that the recognition accuracy of the model is 93%,and it has high accuracy while ensuring real-time performance.
作者
姜晓洁
谷卓琪
柳明顺
杨程云
范蓉洁
JIANG Xiaojie;GU Zhuoqi;LIU Mingshun;YANG Chengyun;FAN Rongjie
出处
《今日自动化》
2024年第9期153-154,157,共3页
Automation Today
