平行坐标法的发动机设计优化数据相关性分析

Motor Multi-variables Optimization Data Relativity Analysis Based on Parallel Coordinates

为有效利用发动机设计分析和优化计算中产生的多变量数据,应用平行坐标法对数据进行可视化。首先将设计方案点映射为平行坐标系中的一系列多边折线,然后采用一定赋色原则解决大量数据点引起的图形重叠问题,并采用一定交互手段选择不同的设计变量、目标函数和约束,最终根据平行坐标与笛卡尔坐标的映射关系,结合各坐标轴上颜色变化趋势分析各设计变量、目标函数和约束之间的相关关系。该方法将多变量相关关系分析问题转变为二维模式识别问题,为设计优化提供一种直观有效的辅助分析手段。某型发动机设计优化数据相关性分析实践证明了该法的有效性。

To make use of motor multi-variables design and optimization data,parallel coordinates methods was presented to visualize them.First design scheme points were mapped to series of polygonal lines.Then the lines were endowed with color in some principle to solve their overlapping.And one interaction means was applied to select different design variables,objective and restrictions.At last,relativities between design variables,objective and restrictions were gotten according to mapping relationship between parallel coordinates and Cartesian coordinates,combined with color changing trend in every variable parallel coordinate.Thus multi-variables relativity analysis problem was translated into 2D pattern identification.It is good and intuitive for assistant design analysis.Practice to apply it to some motor multi-variables optimization data relativity analysis proved its validity.