基于RBF神经网络的裂缝开度预测模型

Prediction Model of Dam Crack Opening Based on RBF Neural Network

大坝裂缝开度是大坝安全的重要特征之一。针对传统裂缝开度预测模型的不足,提出了基于RBF神经网络的裂缝开度预测模型,该模型克服了传统模型容易陷入局部极小和运算迭代量大的缺点,有效地提高学习速度,加快了收敛速度,缩短了训练时间,使得预测精度提高,因此能较好地预测裂缝开度值。利用Matlab的RBF神经网络工具箱建立了裂缝开度预测模型,并应用于实际工程中,取得了较高的拟合预报精度,说明该方法具有较强的实用性。

The dam crack opening is one of the most important characteristics of the dam safety. According to the deficiency of traditional prediction model, the prediction model of dam crack opening based on RBF neural network is proposed. This model can overcome the shortcoming of bogging down in partial minimum and large amount of iterative computation volume. It can quicken the train; improve the learning speed and convergence rate. While it can get higher prediction precision and need much fewer training time. So the model based on RBF neural network can predict the crack opening preferably. In this paper, the prediction model of dam crack opening is established by using the neural networks toolbox of Matlab. This model is applied in practical projects, and it has gotten the higher fitting and forecast accuracy, so this method has strong practicality and high efficiency.