论文部分内容阅读
针对多目标优化设计中存在寻优空间小、计算难度大等问题,提出4维可视化寻优方法,将颜色作为第4变量引入3维坐标系中,分别用三维坐标表示待求自变量,用三维空间诸点的颜色表示目标函数的值。视图中多个切片的同色线集成为空间等值域,获得清晰的4维显性可视化效果,并据图中的交域颜色分布实现多目标寻优。以对逆变器的非对称无源T型低通滤波器进行优化设计为例对该方法进行了详细说明和论证,在扩大的寻优范围内,目标函数的分布状态清晰可见,有效地提高了优化解的稳定性。本方法表明:在寻优过程中应充分考虑目标函数对子变量的敏感性。提供了数学证明,介绍了使用本方法的设计准则,以及用于定量比对的实验数据。
Aiming at the problems of small optimization space and high computational complexity in multi-objective optimization design, a 4-D visualization optimization method is proposed. The color is introduced into the 3-D coordinate system as the 4th variable. The 3-D coordinates are used to represent the independent variables The color of points in 3D represents the value of the objective function. The same color line of multiple slices in the view is integrated into the spatial equivalent region to obtain a clear 4-D explicit visualization effect, and multi-objective optimization is achieved according to the color distribution of the cross-region in the figure. Taking the asymmetric passive T-type low-pass filter of inverter as an example, this method is illustrated and demonstrated in detail. Within the scope of the expanded optimization, the distribution of the objective function can be clearly seen and effectively improved Optimal solution of the stability. The method shows that the sensitivity of the objective function to the sub-variables should be fully considered in the optimization process. Provided mathematical proofs, introduced design guidelines for using the method, and experimental data for quantitative alignments.