摘要
针对传统可视化技术处理高维多目标设计优化问题的不足,对超径向可视化技术开展了研究.该技术将超空间Pareto前沿无损地显示于二维空间,有助于设计人员直观了解复杂的Pareto解集空间、并快速获得较好的权衡设计.此外,在超径向可视化过程中,通过采用权重因子和基于偏好区间的颜色标记原则引入设计偏好,可以辅助设计者选择出更加符合特定偏好的设计方案.将超径向可视化技术应用于近地空间火箭的多目标设计优化,验证了该技术处理高维多目标设计优化问题的可行性和高效性.
Considering the deficiency of traditional visualization methods when dealing with high-dimensional multi-objective design optimization,the hyper-radial visualization(HRV) method was studied.HRV provided a lossless visualization way to represent the hyperspace Pareto frontier,enabled designers to realize the complex Pareto set intuitively,and identified a better trade-off design quickly as well.Besides,with the purpose of assisting designers to choose a more specific optimal design solution,the designers’ preference was incorporated by the weights and color marks with range-based preferences.A multi-objective optimization design of a sub-orbit launch vehicle was accomplished by HRV,which indicates the HRV method is feasible and efficient for high-dimensional multi-objective design optimization.
出处
《航空动力学报》
EI
CAS
CSCD
北大核心
2013年第3期591-596,共6页
Journal of Aerospace Power
关键词
超径向可视化
偏好引入
多目标优化
PARETO解集
固液火箭发动机
hyper-radial visualization(HRV)
preference incorporation
multi-objective optimization
Pareto set
hybrid rocket motor