离散化
序列(生物学)
流量(数学)
采样(信号处理)
过程(计算)
数学优化
启发式
控制理论(社会学)
惩罚法
工程类
算法
计算机科学
数学
机械
物理
化学
电气工程
数学分析
控制(管理)
生物化学
滤波器(信号处理)
人工智能
操作系统
作者
Huan Xia,Dong Xiang,Peng Mou
出处
期刊:Applied Mechanics and Materials
[Trans Tech Publications, Ltd.]
日期:2014-01-01
卷期号:496-500: 697-702
被引量:1
标识
DOI:10.4028/www.scientific.net/amm.496-500.697
摘要
The design of a vector showerhead in a vertical reactor involves thousands of holes on the showerhead face plate and mass transport in the reactor, and the design variables and objective are always spatial related continuous characteristic, so the discretized sequence of the design variables has too many degrees of freedom that the needed sampling points of a design of experiment or heuristic algorithm methods are hardly acceptable. In order to solve this problem, a profile error feedback (PEF) optimization method was proposed. This method employed a loop-iteration-approximation mode, and started from a guessed initial model, and then fed back the profile error of design objective, according to a weight function, to continually adjusted the design variable sequence to make the gas flow uniformity in the new model continually approximate the perfect one. The optimization case of a vector showerhead system shown that the non-uniformity of the gas flow was reduced from 0.36% to 0.03% under a typical process condition.
科研通智能强力驱动
Strongly Powered by AbleSci AI