非线性耦合模型在桥梁桩基沉降预测中的应用研究

摘要

Through the settlement prediction of the bridge pile foundation,it can effectively evaluate and judge the stability of the bridge,providing some guidance for the site construction.At the same time,the systematic prediction methods can effectively improve the prediction accuracy,therefore,the gray model and BP neural network are coupled,establishing the preliminary prediction model for the settlement of the bridge pile foundation,and then use the Markov chain to establish error correction model to implement the phased prediction of settlement.This model has the advantages of grey model " accumulated generating" grey sequences,increases the regularity of deformation data and make full use of the nonlinear prediction ability of BP neural network and Markov chain,with advantages of strong systematicness and high comprehensiveness.At the same time,2 examples are used to verify the results.The results show that the measured values are in good agreement with the predicted values.Among them,the average relative error of example 1 is 1.37 ％,and the average relative error of example 2 is 1.39 ％.The prediction results of two examples have little difference,with high prediction accuracy,validating the effectiveness of this prediction model.%通过对桥梁桩基的沉降预测,能有效地评价和判断桥梁的稳定性,为现场施工提供一定的指导依据.同时,系统性的预测方法能有效提高预测精度,因此,将灰色模型和BP神经网络进行耦合,建立了桥梁桩基沉降的初步预测模型,再利用马尔科夫链建立误差修正模型,实现桥梁桩基沉降的分阶段预测.该模型发挥了灰色模型“累加生成”灰色序列的优点,增加了沉降数据的规律性,又充分利用了BP神经网络和马尔科夫链的非线性预测能力,具有系统性强、全面性高等优点.同时,利用2个实例进行验证,结果表明实测值和预测值较吻合.其中,实例1平均相对误差为1.37％,实例2的平均相对误差为1.39％,两实例的预测结果差异不大,具有较高的预测精度,验证了所提预测模型的有效性.