Displacement prediction of landslide based on new intelligent algorithm of ELM
Considering slow learning speed and complex selection of network structural parameters of conventional intelligent algorithm in landslide displacement prediction, a prediction model for landslide displacement based on Extreme Learning Machine (ELM) is presented in this paper.The number of optimum neurons on hidden layer and excitation function of ELM are determined according to the 2D range search algorithm and the technique of rolling modeling is adopted in prediction in order to improve the network generalization ability and prediction accuracy.Finally, taking Lianziya landslide and Gushuwu landslide as the case, a comparative study was carried out between ELM models with conventional algorithms like LMBP and RBF respectively.The results show that the ELM algorithm has higher accuracy and better network learning speed.
ELMneurons on hidden layerexcitation functionlandslide displacement
国家科技期刊平台
NSTL
万方数据
维普
