基于小波神经网络组合模型的高填方路基沉降预测

Combined forecasting model based on wavelet neural network for prediction of high fill subgrade settlement

利用小波变换良好的时频局域化性质和神经网络的自学习功能,结合S型成长曲线模型,建立了基于小波神经网络的高速公路高填方路基沉降预测模型,该模型的应用避免了计算过程中各种人为因素的影响.通过对汝（城）郴（州）高速公路K59＋375~K59＋445高路堤沉降现场监测数据的学习、预测与检验,并与S型成长曲线模型和BP神经网络的预测结果相比较,结果表明,组合模型的预测精度高,与实际情况相吻合.

Using the nature of good time-frequency localization of wavelet transform and the self-learning function of neural networks,a combined freeway high filling subgrade settlement prediction model with S-Growth Model is established,based on wavelet neural network.The model is based on Real-time monitoring data,avoiding the calculation of various human factors.Through Ru-Chen freeway K59＋375～K59＋445 embankment settlement on-site monitoring data of high learning,prediction and testing,and with S-Growth Model and BP neural network prediction,results show that combined model of wavelet neural network prediction are accurate and consistent with the actual situation.