论文部分内容阅读
将小波神经网络组合预测模型引入软土路基沉降预测中。把5组不同形式的s型增长模型单项预测结果作为小波网络的输入向量,将代表相应时刻的实际值作为小波网络的输出,对软基沉降序列进行非线性组合预测。预测结果表明,小波网络组合预测的结果比各单项模型预测的结果都好,与BP神经网络相比,小波网络的收敛速度更快,预测精度更高,模型的泛化能力更强。
The combination of wavelet neural network prediction model into the soft soil subgrade settlement forecast. Five groups of s-type growth model single prediction results are taken as the input vector of wavelet network, and the actual value of the corresponding time is taken as the output of the wavelet network to predict the nonlinear settlement of the soft foundation settlement sequence. The results show that the results of wavelet network combination prediction are better than those of single model prediction. Compared with BP neural network, wavelet network has faster convergence rate, higher prediction accuracy and stronger generalization ability.