基于遗传算法的单桩极限承载力灰色预测法

在电厂建设工程中大承载力桩非常普遍,此类桩在静载试验中不能达到极限状态或破坏阶段,其极限承载力通常依据有限实测数据通过计算进行预测,灰色预测法便是其中常用的一种方法.传统GM(1,1)灰色预测模型的拟合程度与预测精度有限,本文结合灰色预测的检验方法,将后验差比作为目标函数,实现了基于遗传算法的单桩极限承载力灰色预测法.应用实例表明,与传统灰色预测法相比,该方法能有效减小预测结果的后验方差比,并能得出更合理的桩基极限承载力.

Gray Prediction ultimate bearing capacity of single pile based on genetic algorithm

Piles with huge bearing capacity are common in power plant construction engineering.These piles can't reach the ultimate status or the fail stage in static test,so the ultimate bearing capacities are usually predicted by computation based on limited measured values;for example,the gray prediction method.The fitting effect and prediction accuracy of traditional GM(1,1) gray prediction model are limited.The gray prediction based on genetic algorithm for determining ultimate bearing capacity of single pile is realized with the posterior-variance-ratio as objective function.The applications show that comparing to traditional gray prediction,this method can effectively decrease the posterior-variance-ratio of prediction results,and can get more appropriate ultimate bearing capacity of single pile.