基于混合遗传-Nelder Mead单纯形算法的源强及位置反算  

Back-calculation of source strength and position by a hybrid genetic-Nelder Mead simplex algorithm

作　　者：张建文[1,2] 王煜薇[1] 郑小平[1] 王正[2] 

机构地区：[1]北京化工大学安全管理研究所,北京100029 [2]中国石油化工股份有限公司化学品安全控制国家重点实验室,青岛266071

出　　处：《系统工程理论与实践》2011年第8期1581-1587,共7页Systems Engineering-Theory & Practice

基　　金：北京市优秀人才培养资助(ZG2007-68);化学品安全控制国家重点实验室开放研究基金和国家科技支撑计划(2006BAK01B03)

摘　　要：确定泄漏源的位置和强度,是进行群体疏散和应急决策的基础.将扩散模型得到的浓度值与传感器观测的浓度值进行比较并建立混合遗传-Nelder Mead单纯形算法模型,反算得到泄露源的位置和强度,进而利用浓度的模拟数据验证该算法的可行性.研究结果表明:混合遗传-NelderMead单纯形算法不受初值选取的影响,即使初值远离期望值,也能得到很好的结果,而且能以较小的误差和较快的速度反算出结果,更适合于多维变量的搜索.因此混合遗传-Nelder Mead单纯形算法能够快速准确地反算得到泄漏源的位置和强度,满足应急决策的需要.Determination of the locations and strength of the source in a chemical leakage is crucial to crowd evacuation and emergency decision.This paper compares concentrations computed by dispersion model with the measured by receptors and then hybrid genet ic-Nelder Mead simplex algorithm model is established.The locations and strength of leakage source are characterized.The feasibility of the algorithm is verified by simulation data.It is indicated that the hybrid genetic-Nelder Mead simplex algorithm is seldom affected by initial values and can get good results even if the initial values are far away from the expected values.This method back-calculates the optimal solution in a smaller errors and a faster pace and is more suitable for the search procedure in multi-dimensional space.Thence,hybrid genetic-Nelder Mead simplex algorithm is able to back-calculate the locations and strength of leakage source quickly and accurately,which will meet the needs of emergency rescue.

关 键 词：混合遗传-Nelder Mead单纯形算法 源强反算 初值 误差 多维变量 

分 类 号：X928.5[环境科学与工程—安全科学]