基于改进遗传-单纯形混合算法的危险气体泄漏溯源分析

摘要

When the dangerous gas leakage occurs,it is the basis and foundation for making the emergency plan to determine the location and source intensity of leakage.When it is unable to directly determine the leakage location and measure the source intensity of leakage,it is necessary to inverse the possible location and source intensity of leakage through the several limited monitoring points,but the existing methods have many disadvantages,such as too slow convergence,sensitive initial value,too many parameters,etc.An IGA-NM hybrid algorithm combined with improved genetic algorithm and simplex method was put forward to quickly calculate the location and source intensity of gas leakage.The IGA-NM hybrid algorithm not only avoids the too slow convergence of GA,but also avoids the sensitive initial value of NM,and takes into account the global optimization.Compared with GA and NM,the IGA-NM hybrid algorithm is faster in computation speed and smaller in computation error.Based on the IGA-NM hybrid algorithm,a computer program was designed based on webGIS to calculate the source intensity and location of gas leakage,which simplifies the input parameters,being easy to use,and can be applied to emergency monitoring of gas leakage,traceability reversion of air pollution source and other occasions.%当发生危险气体泄漏时,确定其泄漏位置和泄漏源强,是制定应急方案的基础和依据之一.当无法直接确定泄漏位置、测量泄漏源强时,就需要通过有限的几个监测点,反演出可能的泄漏位置和泄漏源强,现有方法存在收敛过慢、初值敏感、参数过多等问题.描述一种结合改进遗传算法和单纯形法的IGA-NM混合算法,可用于快速反算气体泄漏的位置和源强.IGA-NM混合算法既避免了GA的收敛过慢,又避免了NM初值敏感,兼顾了全局优化.与GA、NM相比,IGA-NM混合算法的计算速度更快,计算误差更小.最后,应用IGA-NM混合算法,基于WebGIS设计了一套计算气体泄漏源强和位置的计算机程序,简化了输入参数,使用方便,可适用于气体泄漏应急监测、大气污染源溯源反查等场合.