优化反演方法在工程反演计算过程中,需要进行反复的正演计算,导致效率非常低。本文提出了目标函数的改进形式,并完成了改进形式的计算求解推导,大大减少了优化反演算法的正演算次数,提高了计算效率。在此基础上,与传统遗传算法相结合,开发了线弹性多参数优化反演计算程序。以小湾水电站为工程实例,反演了施工期坝基分区综合变形模量。结果表明,所提出的改进方法克服了传统遗传算法需要大量正演计算的固有瓶颈,极大地提高了反演计算效率,且计算精度可靠。改进后的目标函数同样可以与其它优化算法相结合开发相应的改进优化反演程序。
In engineering practice the optimized back analysis needs a lot of repeated computation and results in low efficiency of computation. An improved objective function is proposed to solve this problem. The solution of this function is deduced. By using this function the amount of computation number can be remarkably reduced. On this basis, combining with the traditional genetic algorithm, the computational program for optimized back analysis of linear elastic multiple parameters is developed. The proposed method is applied to back analysis of the deformation modulus of the dam foundation in the Xiaowan Hydropower Project during excavation period. The result shows that the computation efficiency is greatly elevated.