节点文献

基于深度图像的加工特征识别算法在模具数控智能编程中的应用

Application of machining feature recognition algorithm in die NC intelligent programming based on depth image

  • 推荐 CAJ下载
  • PDF下载
  • 不支持迅雷等下载工具,请取消加速工具后下载。

【作者】 孙志晖; 朱凌穹; 王华昌; 李建军;

【Author】 SUN Zhihui;ZHU Lingqiong;WANG Huachang;LI Jianjun;State Key Laboratory of Material Processing and Die & Mould Technology, Huazhong University of Science and Technology;Wuhan Eman Technology Co., Ltd.;Hubei Huangshi Mold Industry Technology Research Institute;

【机构】 华中科技大学材料成形及模具技术国家重点实验室; 武汉益模科技股份有限公司; 湖北黄石模具产业技术研究院;

【摘要】 针对传统的基于规则推理的特征识别算法灵活性低及基于卷积神经网络的三维模型特征识别算法计算量大的问题,提出一种基于深度图像的加工特征识别算法。该算法通过提取三维几何模型的深度图像,使用基于深度学习的目标检测算法获取二维深度图像中的加工特征信息,利用NX/Open API和NVIDIA推出的通用并行计算架构在UG NX12.0平台上开发了基于该方法的原型系统,在由数千个注射模电极零件构成的模型库中,随机选取部分零件进行验证。结果表明,针对电极零件的典型加工特征,识别正确率达到90%以上,该算法提高了加工特征识别的灵活性与效率。

【Abstract】 Aiming at the problems of low flexibility of the traditional feature recognition algorithm based on rule reasoning and large computation of the 3D model feature recognition algorithm based on convolution neural network, a processing feature recognition algorithm based on depth image was proposed. The algorithm extracted the depth image of the 3D geometric model, and used the target detection algorithm based on depth learning to obtain the processing feature information in the 2D depth image. A prototype system based on this method was developed on the UG NX12.0 platform using the NX/Open API and the compute unified device architecture launched by NVIDIA. Some parts were randomly selected from the model library composed of thousands of injection mould electrode parts for verification. The results showed that the recognition accuracy of typical machining features of electrode parts could reach more than 90%, and the algorithm effectively improved the flexibility and efficiency of machining feature recognition.

  • 【文献出处】 模具工业 ,Die & Mould Industry , 编辑部邮箱 ,2023年07期
  • 【分类号】TP391.41;TG76;TG659
  • 【下载频次】29
节点文献中: 

本文链接的文献网络图示:

本文的引文网络