边坡位移预测的RBF神经网络方法

利用边坡实测位移序列来预测边坡未来时间的位移,可以有效地判断边坡的稳定性。由于神经网络可以通过对样本的反复学习来反映边坡复杂的非线性演化关系,其预测效果要优于传统的预测方法。RBF神经网络作为一种性能良好的前馈网络,具有更好的逼近能力和全局最优特性。以边坡位移时间序列为基础,采用RBF神经网络建立边坡位移预测模型,通过最近邻聚类学习算法实现边坡位移预测,具有结构简单、学习速度快、预测精度高的特点,网络的外推能力也较强。通过2个工程实例说明边坡位移预测的RBF神经网络方法的有效性。

METHOD TO FORECAST DISPLACEMENT OF SLOPE BASED ON RBF NEURAL NETWORK

Based on the displacement sequence of slope,the stability of slope could be judged effectively by forecasting the displacement of slope in the future.Because the nonlinear evolution of slope could be obtained by learning the sample repeatedly with neural network,the forecast effect of neural network is better than those of traditional methods.Compared with BP neural network,RBF neural network has better capability of approximation and globe optimum characteristics as a well-behaved feedforward network.Based on the time series of the slope displacement,RBF neural network is adopted to construct the forecast models;and the nearest neighbor-clustering algorithm is used to forecast the displacement of slope.The method has the advantages of simple structure,faster learning velocity and better precision of forecast;and the extrapolated capability of RBF neural network is better than that of BP neural network.Finally,two engineering examples are given to testify the effectiveness of the forecast method to displacement of slope based on RBF neural network.