基于NSGA-Ⅱ算法的堆石坝多目标参数反演方法

发布时间：2018-05-30 18:47

本文选题：堆石坝 + 参数反演　；参考：《水力发电学报》2017年01期

【摘要】：参数反演确定堆石体的强度变形参数已成为堆石体参数取值的有效途径之一。在目前的堆石体参数反演方法中,未考虑不同材料分区堆石体力学特性的差异和各自分区的应力变形特点,对整个坝体或某个断面采用一个目标函数进行参数反演。本文考虑堆石坝不同材料分区的受力变形特点,建立不同材料分区的反演目标函数,并基于非支配排序的多目标遗传算法(NSGA-Ⅱ),兼顾各分区间材料特性对变形的相互影响,提出了基于NSGA-Ⅱ和RBF神经网络的堆石坝多目标参数反演方法。以水布垭堆石坝为例,采用该方法对坝体主堆石和次堆石区的瞬变-流变参数进行了反演分析,并与单目标参数反演方法进行了对比分析。计算结果表明,所有测点的变形量和变形趋势上与实测值基本一致,反演效果明显优于单目标参数反演结果,表明该方法对于多分区堆石坝的材料参数反演更为合理有效。
[Abstract]:Parameter inversion to determine the strength and deformation parameters of rockmass has become one of the effective ways to determine the parameters of rockfill parameters. In the current inversion method of rockfill parameters, the differences in the mechanical properties of different materials and the stress and deformation characteristics of the respective zoning are not taken into account, and a target function is adopted for the whole dam body or a section. In this paper, we consider the characteristics of the stress and deformation of different material zoning of the rockfill dam, establish the inverse objective function of different material zoning, and based on the multiobjective genetic algorithm (NSGA- II) of non dominated sorting (NSGA- II), take into account the interaction between the material characteristics of each partition, and propose a multi target parameter of the rockfill dam based on the NSGA- II and the RBF neural network. Taking the Shuibuya rockfill dam as an example, this method is used to analyze the transient and rheological parameters of the main rock and the secondary rock area of the dam, and compared with the single object parameter inversion method. The results show that the deformation and deformation trend of all the measured points are basically consistent with the measured values, and the results of the inversion are obviously superior. The result of single target parameter inversion shows that the method is more reasonable and effective for the inversion of material parameters of multi zone rock fill dams.
【作者单位】：武汉大学水资源与水电工程科学国家重点实验室;长江勘测规划设计研究院;
【基金】：国家自然科学基金(51579193;5132290;51509190) 中国博士后科学基金面上资助(2016T907272)
【分类号】：TV641.43

【相似文献】