基于BP神经网络-马尔科夫链模型的隧道围岩位移预测

在隧道工程施工中，围岩位移预测起着很重要的作用。将BP神经网络-马尔科夫链模型引入到隧道围岩位移预测中来，通过对训练样本的学习，利用BP神经网络实现了对位移时间序列的滚动预测，同时得到了实测值与预测值的相对误差；在此基础上利用马尔科夫链对相对误差进行修正，有效地提高了预测结果的精度。并将该模型应用于某公路隧道拱顶下沉位移时序预测中，结果表明该模型具有精度高、科学可靠的特点，为隧道围岩变形的预测提供了新的途径。

Forecast of Tunnel' s Surrounding Rock Displacement by BP Neural Network and Markov Chain

Forecast of surrounding rock displacement is significant for tunnel engineering. The model of BP neural network Markov chain was adopted to the displacement forecast for tunnel surrounding rock. Through emulating the training samples, rolling forecast for the displacement time series was performed by BP neural network, and the relative error of measured and predicted values was acquired. Furthermore, the Markov chain was employed to correct the relative error, and  the forecast results were improved. The model was applied to the time-series forecast of the vault settlement of a real vehicular tunnel, and the result showed that the model is of high precision and reliability. It provides a new approach for the forecast of tunnel's  surrounding rock displacement.