复杂零件潜在缺陷识别与再设计靶向抑制技术研究

Potential defects inside the complex machine part have strong concealment. Identify potential defects and then inhibit the duplication by using process redesign can improve the product design and manufacturing quality, which is of great significance. The contents of the research project include: potential defect detection model of complex parts, potential defect classification under non-uniform interference , target mapping between typical potential defects and process design , and potential defects target suppression method for redesign. The contributions of the research project are: potential defect quantitative detection technology based on the hybrid two-mode group-point-pair representation model , potential defect classification method based on the multi-feature-space and importance degree distribution, and potential defects target redesign suppression technology based on the precedence constraint matrix. Based on theoretical studies, develope the potential defect identification and redesigning targeted inhibition system, and take the closed blisk as the application object to verify the proposed method. The completion of the project has vast importance to ensure the safety running of mechanical equipment under multi-interference.

复杂零件潜在缺陷具有很强的隐蔽性，识别潜在缺陷并通过工艺再设计抑制其重复产生，对提高产品设计与制造质量具有重要意义。项目研究内容包括复杂零件潜在缺陷影像表征模型、不均匀干扰下潜在缺陷分类识别、典型潜在缺陷与工艺耦合靶向映射、以及面向再设计的潜在缺陷抑制方法。在基于双模式混合点对群表征模型的潜在缺陷目标辨识方法、基于多特征空间及属性重要度分配的潜在缺陷分类识别方法、以及基于优先约束关联矩阵的潜在缺陷工艺再设计靶向抑制技术方面取得创新。开发潜在缺陷识别及再设计靶向抑制原型系统，并以闭式叶盘类零件为应用对象进行应用验证。本项目的完成，对保障多干扰因素下机械零件设计与制造质量有着重要的理论及应用价值。

机械零部件在制造、安装和使用过程中不可避免地存在各种缺陷。本项目在潜在缺陷检测、识别以及再设计缺陷抑制方面取得了一系列重要成果。（1）提出了基于多模式点对群的潜在缺陷表征模型构建及缺陷检测方法。建立了具有统计特性的多模式混合点对群表征模型，实现了含复杂结构的潜在缺陷精确检测。（2）提出了基于随机三角分布及稀疏表示分类的潜在缺陷识别方法。采用由缺陷形状形成的随机三角采样来产生连续的概率分布，可在旋转、不同尺度以及噪声情况下对不同类型缺陷进行分类识别。（3）提出了考虑sobol缺陷敏感度的抗恶劣环境结构均衡设计方法。基于抽象缺陷模型，模拟缺陷对结构的侵袭及迁移过程，有选择性的选取优化变量进行优化再设计，提高零部件结构敏感部位的安全度及抗缺陷性能。本项目进行期间相关研究成果在国际上具有较大的影响力，共发表了SCI/EI论文8篇（SCI检索论文6篇，EI检索论文2篇），授权发明专利2项，申请发明专利4项。对提高复杂机械零件设计与制造质量具有重要意义。