Risk assessment of landslides along the Yizhan railway based on the CF-BPNN coupling model
Landslide is a common geological hazard that can easily cause serious casualties and property damage,especially the series of landslides developed along the railway line,which pose great risks and hidden dangers to lifeline engineering.This article takes the Yiyang to Loudi section of the Yizhan railway as the research object.Based on the geological disaster results of the first natural disaster risk survey and field geological survey data,16 types of geological environmental factors are extracted from 4 aspects:topography,regional geology,hydrogeology,and human activities to construct a landslide risk assessment system.The deterministic coefficient model(CF)is introduced to improve the traditional BP neural network model(BPNN),and landslide risk assessment is carried out to enhance the performance of the BPNN model and improve the accuracy of prediction.On this basis,summarize the distribution characteristics of landslides in the study area,among which elevation,lithology and roads have important impacts on the distribution of landslides in the study area.Finally,the improved coupling model was compared and analyzed with a single CF model and BPNN model through ROC curves.The results show that the AUC value of the CF-BPNN model is 0.849,the AUC value of the CF model is 0.754,and the AUC value of the BPNN model is 0.837,which indicates that the improved coupling model is better than the single model,and the prediction accuracy is higher.The research results can provide information support for landslide risk analysis of near-field lifeline engineering.
万方数据
维普
