基于小波神经网络的隧道变形预测模型研究

摘要

Based on the fact that single model’s prediction precision in tunnel deformation monito-ring is not very high,this paper proposes a Neural Network model based on Wavelet analysis theory, which can overcome slow convergence speed,blindness of structure design and local minimum of BP Neural network.Compared with time series model,Levenberg-Marquardt BP Neural Network model,and GA-BP Neural Network model,prediction results show that Wavelet Neural Network model has a simple network structure,faster convergence speed and higher prediction precision in tunnel deformation monito-ring.%针对单一模型在隧道变形预测上精度不高的问题，提出了一种基于小波分析理论的神经网络模型，该模型克服了 BP神经网络模型存在的收敛速度慢、结构设计盲目、易陷入局部极小点的缺陷，通过将该模型与时间序列模型、Levenberg-Marquardt法 BP神经网络模型、遗传神经网络模型预测的结果比较，可以看出小波神经网络在隧道的变形预测中网络结构更简单、收敛速度更快、预测精度更高。