摘要
在参数设计中,迄今所介绍的各种寻优方法,都建立在逐轮正交试验基础上,均未能解决搜索求解的方向问题。该文应用梯度理论建立的SN比梯度搜索水平迭代法,明确给出了试验结果,迅速改善的搜索寻优方向,即SN比梯度方向,并给定了沿该方向逐步进行搜索迭代的具体算法。计算机实例验证证实,该法寻优方向明确、效率高、结果好。
In parameter designs,all of current approaches are based on making repeatedorthogonal experiment and have not solved the problem of searching directions。On the basis of gradient theory,this paper presents a new approach that points out clearly searchingand solving directions,or SN ratio gradient direction,in which searching effect can bequickly improved.Additionally,the algorithm of the approach is given。 The approach ischaracterised by its clear searching diretion, high searching efficiency and good searchingeffect.
出处
《南京理工大学学报》
CAS
CSCD
1995年第4期358-361,共4页
Journal of Nanjing University of Science and Technology
关键词
迭代法
试验设计
SN比梯度
参数设计
iteration method
experimental designs
signal noise ratio gradients