CAJ | 학술논문

考虑到软土的复杂性和各种沉降理论计算方法的局限性，利用有限的沉降实测数据，选取合理的模型及方法来预测软基的沉降。以深圳盐田港西港区纳泥塘地区软基沉降预测为例，采用了BP向前型网络模型和Elman反馈型网络模型两种不同的神经网络建模法，通过高度非线性的曲线拟合，推求工程后期沉降(包括最终沉降)，并与曲线拟合法中的双曲线法、指数曲线法、泊松曲线法及 Asaoka 法对比，对拟合预测结果进行检验，使其具有统一的量化标准。对比结果表明：BP神经网络模型和双曲线法的预测效果最好，适用于本工程的沉降预测。
고필도연토적복잡성화각충침강이론계산방법적국한성，이용유한적침강실측수거，선취합리적모형급방법래예측연기적침강。이심수염전항서항구납니당지구연기침강예측위례，채용료BP향전형망락모형화Elman반궤형망락모형량충불동적신경망락건모법，통과고도비선성적곡선의합，추구공정후기침강(포괄최종침강)，병여곡선의합법중적쌍곡선법、지수곡선법、박송곡선법급 Asaoka 법대비，대의합예측결과진행검험，사기구유통일적양화표준。대비결과표명：BP신경망락모형화쌍곡선법적예측효과최호，괄용우본공정적침강예측。
Due to the complicated character of soft soil and the limitation of the present settlement calculation methods,the settlement is predicted by the predicted model and methodology from the actual settlement data.The settlement prediction in soft founda-tion treatment of Nanitang in Yantian port of west port area of Shenzhen is taken as an example,two different modeling methods of the neural network including BP and Elman are used for different conditions and requirements in practical engineering,then the late settlement(including the final settlement)can be predicted by highly nonlinear curvefit-ting.The methods of curve fitting method such as hyperbolic method,exponential meth-od,poission curve method and Asaoka method are verified and contrasted,the check is used to discuss the results of fitting and prediction so that all the methods can be evalua-ted by using an quantized standards.The results show that BP and the hyperbolic method are considered as the most effective ones among methods and can be adapted to predict the settlement of this engineering.