基于GA-GM的砂性富水地层深基坑地表沉降预测

Ground Settlement Prediction of Deep Foundation Pit in Sandy Water-Rich Stratum Based on GA-GM

深基坑施工会引起沉降,对高度非线性的沉降数据进行研究可预先判断发展趋势,是保障基坑安全的有效方法之一.根据传统灰色预测模型(GM)得到α、b的初始值及搜索域,基于MATLAB利用遗传算法(GA)对GM的α、b进行参数寻优.结合某工程实例,结果表明:GA-GM模型在围护阶段DBC14-1的α值为-0.0113,b值为6.4404,最终预测与实际结果的相对误差分别为0.0007%、4.5%和4.6%;在施工阶段DBC14-1的α值为-0.0019,b值为21.3629,最终预测与实际结果的相对误差分别为0.505%、0.761%和0.63%,并利用GA-GM对其他测点进行预测,发现与实际值相差较小;在与其他预测模型对比中,发现GA-GM对基坑沉降的预测更具有适用性及准确性.

Construction of deep foundation pits will cause settlement.Studying the highly nonlinear settlement data can predict the development trend,making it one of the effective methods to ensure the safety of foundation pits.Based on the traditional grey prediction model(GM),the initial values ofαand b and the search domain are obtained,and the parameters ofαand b of GM are optimized on MATLAB using genetic algorithm(GA).Combined with an engineering example,the results show that theαvalue of the GA-GM model DBC14-1 in the enclosure stage is-0.0113,the b value is 6.4404,and the relative errors of the final prediction and actual results are 0.0007%,4.5%and 4.6%respectively;during the construction phase,theαvalue of DBC14-1 is-0.0019,and the b value is 21.3629,and the relative errors of the final prediction and actual results are 0.505%,0.761%,and 0.63%,respectively.When the GA-GM method is used to make predictions at other measurement points,the differences between the prediction results with the actual values are also small.It is found that in comparison with other prediction models the GA-GM model has wider applicability and more accuracy in the prediction of foundation pit settlements.