基于BP神经网络修正的自适应灰色模型的隧道变形预测研究

隧道围岩具有高度的非线性变形特征,通过变形预测能有效判断隧道变形的发展趋势。首先以自适应GM(1,1)模型对隧道变形进行初步预测,且为保证自适应模型的参数为全局最优参数,提出以粒子群算法对模型参数进行优化;其次,以BP神经网络为基础,建立误差修正模型,旨在进一步提高预测精度。在此基础上,将该预测模型应用于2个工程实例中,结果表明:该预测模型在横向和纵向上的预测效果均较好,自适应能力和递推能力均较强,预测结果与实测值较为吻合,预测精度较高,能较好地反映隧道围岩的变形规律。该预测模型能较为有效地实现隧道围岩的动态预测,可以进行推广应用及研究,为隧道变形预测提供一种新的思路。

Research on Deformation Prediction of Tunnel Based on Adaptive Grey Model of BP Neural Network Correction

The tunnel surrounding rock has the characteristics of high nonlinear deformation,which can be used to identify the development trend of tunnel deformation. In this paper,tunnel deformation is predicted preliminarily with adaptive GM( 1,1) model,and the model parameters are optimized by particle swarm algorithm to ensure that the parameters of the adaptive model are the global optimal;secondly,the error correction model is established based on BP neural network to further improve the prediction accuracy. On this basis,the prediction model is applied to two engineering cases. The results show that the prediction model has better predication results in horizontal and vertical prediction with strong adaptive and recursive abilities. The predication results are proved in good agreement with the measurements with high accuracy to better reflect the deformation law of tunnel surrounding rock. The model can effectively conduct dynamic prediction of tunnel surrounding rock and can be used widely in tunnel deformation prediction.