改进的遗传算法在堆石体参数反演中的应用

Application of improved genetic algorithm to back analyzing parameters of rockfill

在研究堆石坝问题中,堆石体参数的合理估计非常重要,根据工程实测值反演堆石体参数不失为一种有效估计参数的新思路。在传统遗传算法操作过程中,引入模拟退火的 Metropolis 接受准则,并结合系统识别的基本原理改进了遗传算法。将改进的新算法应用于堆石坝主要堆石料的参数反演,结果显示其误差很小、收敛速度快、精度高的优越性,克服了传统的梯度优化方法和单纯形法所具有的搜索速度随反演参数增多呈级数减慢、容易陷入局部极值点和误差传递导致不收敛等缺点,值得在本领域参数优化中推广。

It is very important to reasonably estimate parameters of rockfill in study and analysis of rockfill dams. Back analysis of parameters of rockfill according to in-situ measurement is one kind of new method to effectively estimate the parameters. Metropolis accepting rule of simulated annealing algorithm and system identification are applied to improve genetic algorithm during the processing. Applying this improved genetic algorithm to back analyzing parameters of rockfill, the result shows that the algorithm has advantages of higher accuracy, quick convergence etc.. This method is deserved to be popularized for parameters optimization in this filed.