论文部分内容阅读
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 effectively utilize multivariable data generated in engine design analysis and optimization calculations, data is visualized using parallel coordinates. Firstly, the design scheme is mapped to a series of polygonal lines in the parallel coordinate system. Then, a certain number of color points are used to solve the problem of graphic overlap caused by a large number of data points. Different design variables, objective functions and constraints are selected through some interactive methods. Finally, According to the mapping relationship between parallel coordinates and Cartesian coordinates, the correlation between each design variable, the objective function and the constraints is analyzed according to the trend of color change on each coordinate axis. The method transforms the problem of multivariate correlation analysis into two-dimensional pattern recognition, providing an intuitive and effective auxiliary analysis method for design optimization. The correlation analysis of some engine design optimization data proved the validity of this method.