Faster R-CNN 的
Faster R-CNN 的
复制标题
DOI：
--
复制
发表时间：
2020
期刊：
影响因子：
--
通讯作者：
钟嘉俊;， 贺德强;苗剑 ，;， 陈彦君;， 姚晓阳;Jiajun Zhong;Deqiang He;Miao Jian;Yanjun Chen;Xiaoyang Yao
中科院分区：
文献类型：
--
作者：
钟嘉俊;， 贺德强;苗剑 ，;， 陈彦君;， 姚晓阳;Jiajun Zhong;Deqiang He;Miao Jian;Yanjun Chen;Xiaoyang Yao
焊接缺陷严重威胁着列车运行安全。为解决地铁车辆铝合金车体焊缝的漏检和误检问题，提出一种基于改进Faster R-CNN的方法。首先，利用Abaqus软件对铝合金车身的焊接缺陷进行模拟，得到多组相似缺陷。然后，基于Faster R-CNN框架对缺陷进行分类，并引入Unet模型和Resnet模型对原有Faster R-CNN框架进行改进，提高识别精度。最后，通过对噪声信号图的检测，验证了模型的鲁棒性。仿真结果表明，改进后的模型对铝合金车身焊缝缺陷检测具有较高的识别率和鲁棒性。
: The safety of train operation is seriously threatened by welding defects. In order to solve the problem of missing detection and wrong detection in aluminum alloy body weld of metro vehicles, a method based on improved Faster R-CNN is proposed in this paper. Firstly, the weld defects of aluminum alloy car body were simulated by Abaqus, and several groups of similar defects were obtained. Then, defects are classified based on the Faster R-CNN framework, and Unet model and Resnet model are introduced to improve the original Faster R-CNN framework to improve the recognition accuracy. Finally, the noise signal graph is detected to verify the robustness of the model. The simulation results show that the improved model has a higher recognition rate and robustness for Aluminum car body weld defect detection.