基于小波理论与人工神经网络的滑坡变形预测

滑坡的位移是由多种因素影响的复杂非线性问题。滑坡位移的预测一直是滑坡研究中的重点与难点。位移预测对于分析滑坡的发展趋势具有重要的作用。本文采用小波理论与人工神经网络理论，对滑坡位移进行预测研究。由于滑坡的位移包括一定的偶然因素，如降雨等不定因素，对滑坡位移的预测存在一定的干扰作用。故本文采用小波理论对滑坡位移进行降噪，提取数据中的有用部分。将降噪后的数据采用人工神经网络进行滑坡位移的预测。由数据结果可见，滑坡预测位移与真实位移间的误差为2%～3%，预测结果好，证明本文所用方法能较好的运用到滑坡位移的预测中。

PREDICTION OF THE LANDSLIDE DISPLACEMENT BASED ON THE WAVELET THEORY AND BP NEURAL NETWORK

The displacement of the landslide is affected by many nonlinear factors. The prediction of the landslide displacement, which is important for analyzing the development of landslide, is difficult but important. In this paper, wavelet theory and BP neural network were applied to predict the displacement of the landslide. Since the displacement of the landslide comprises accidental factors, such as rainfall and construction vibration, the prediction of the landslide was disturbed. In this paper, the displacement was de-noised by using wavelet theory, and the useful part was extracted. The data was then used to predict by using BP neural network. According to the result, the error between the predicted data and the real data was located in 2%~3%.It is proved that the methods used in this paper can be successfully used in prediction of the landslide displacement.