基于人工蜂群优化小波神经网络的隧道沉降预测

Prediction of Tunnel Settlements by Optimized Wavelet Neural Network Based on ABC

小波神经网络存在收敛速度较慢、易陷入局部最优的缺陷,而人工蜂群算法收敛速度快且同时具有局部和全局搜索的能力。文章利用人工蜂群算法对小波神经网络进行优化,形成人工蜂群小波神经网络,并将其应用于地铁隧道沉降预测;并以深圳地铁10号线为例,将该模型的预测结果与BP神经网络、小波神经网络进行对比分析。结果表明,人工蜂群小波神经网络较其它两种模型的预测精度更高,预测结果更稳定。

Since the wavelet neural network has the defects of slow convergence rate and proneness of local optimum,while the artificial bee colony algorithm is capable of local and global search,an optimization of wavelet neural network(WNN)was conducted using artificial bee colony(ABC)algorithm to form artificial bee colony-wavelet neural network(ABC-WNN)and it was applied to prediction of settlement of metro tunnels.Taking the Shenzhen metro line 10 for example,a comparison and analysis were carried out regarding the predicted results and that of BP neural network(BPNN)and WNN.The results show that the prediction result of ABC-WNN is more accurate and stable than that of other two models.