基于异构隐式信息动态融合的产品自适应造型方法研究

The massive heterogeneous data is generated from the product digital design. It is difficult but of more significance to exploit more implicit information buried in the data bank. The methods for extracting, recognizing, and fusing the implicit information from massive heterogeneous data, and reusing the implicit information to improve physical properties of the product efficiently and effectively, become the hot topic in the current research. Therefore, this project focuses on the adaptive product modeling methods based on heterogeneous implicit information fusion. First, the implicit information extraction and recognition methods are studied, for the purpose of providing the semantic modeling feature for adaptive shape modeling of mechanical product. Second, the mapping relationship between the operation functions, including distance and velocity field of the level set, and the implicit information fusion process is analyzed. The heterogeneous implicit information is dynamic, integrated into the voxel based on fusion, visual and adaptive modeling. Moreover, the extended finite element method is adopted to analyze the physical properties together with the information fusion procedure, and the stiffness matrix is modified by the radial basis function (RBF) method according to the type of fusing implicit information. The Chebyshev semi-iterative approach is proposed to accelerate the sparse matrix solving of the dynamic integration procedure. Besides, the structure stress sensitive to shape feature combination and optimization method, the streamline-guided feature line edit method, and the geometric constraints based model deformation method are studied, constructing the performance oriented adaptive modeling system. This research aims to provide theoretical guidance for the adaptive product modeling by fusing the heterogeneous implicit information. The research results have important practical significance in regard to improvement of the performance of physical-aware innovative and improve the product design.

产品数字化设计衍生大量异构数据，如何深度挖掘蕴含其中的隐式信息，通过隐式信息动态融合和快速物理性能仿真，构建性能导向的产品自适应造型是当今研究前沿。由此，本项目拟研究基于异构隐式信息动态融合的产品自适应造型方法：研究面向自适应造型的产品异构隐式信息提取和识别方法，发掘自适应造型的特征和语义素材；探求水平集中距离场和速度场操作函数与隐式信息融合的映射关系，根据隐式信息融合建模和可视表达构建，实现隐式信息动态融合的自适应造型；研究扩展有限元分析和水平集编辑相结合方法，设计基于径向基函数的刚度矩阵动态修正算法，采用切比雪夫并行求解，实现自适应造型设计过程中物理性能快速仿真；研究应力感知的造型特征组合，流线引导的特征线编辑和带几何约束的模型变形方法，实现性能导向的造型特征自适应编辑。预期研究成果可为产品异构信息融合和自适应造型设计提供理论指导，并有效提高物理性能感知的产品创新和改进设计水平。

复杂产品设计、制造和测量等各个环节产生大量非结构化、离散化和无序化的模型，累积了大量混杂数字化模型数据，然而庞大的模型数据和信息却难以形成设计知识和设计规律。因此，如何有效利用复杂产品现有海量异构设计数据，构建性能导向的产品自适应设计方法成为研究的重点和难点问题。由此，本项目研究了基于异构隐式信息动态融合的产品自适应造型方法。首先根据复杂产品海量异构模型数据，研究了面向自适应造型的复杂产品异构隐式信息提取和识别方法，获得了自适应造型的特征和语义素材。然后根据提取语义特征信息，研究了基于水平集的距离场和速度场操作函数与隐式信息融合的映射关系，构建了基于移动立方体方法的隐式信息融合建模和统一可视表达，实现了隐式信息动态融合的自适应造型设计。接着，研究了扩展有限元分析和水平集编辑相结合的分析方法，提出了基于径向基函数的刚度矩阵动态修正算法，并研究了基于切比雪夫并行求解的自适应造型设计过程的物理性能快速仿真。最后，研究了基于敏感性分析的物理性能感知的造型特征组合方法，提出基于变密度方法的微结构自适应造型和拓扑优化方法，实现性能导向的造型特征自适应设计和编辑。相关理论研究成果，进一步运用到轨道车辆虚拟限界计算、螺旋路径加工以及轨道交通轮轨磨耗自动检测中，有效提高物理性能感知的产品创新设计水平。项目资助在国内外学术期刊上发表论文12篇，其中高水平SCI论文10篇，中文核心期刊论文2篇；申请了国家发明专利6项，其中授权了4项国家发明专利；参加美国机械工程协会举办的高水平国际学术会议1次，并在分会场做学术报告。项目投入直接经费20万元，支出13.44万元，剩余经费6.56万元将用于本项目研究的后续支出。

内容获取失败，请点击重试