Research on the hazard assessment of regional landslide contributes essential practical significance and study value to disaster preventing and rescuing

land planning

and disaster managing and decision

and so on.The hilly region of Sichuan Province is located in the east of the province

this area is in the subtropical monsoon climate zone

and its abundant rainfalls provide sufficient water conditions for the development of landslides

which make the highest density of landslides in our country.This paper

taking it as the study area

selects 8factors（including height

slope

relief degree of land surface

lithology

fault

road

river

NDVI etc.）as disaster pregnant environmental factors of the landslide with analyzing the pregnant environment in depth based on the landslide catalog database

and then introduces the CF-MLP model by fusion of certainty factor and multiple layer perceptron to assess landslide susceptibility in the study area

the better recognition results to landslide are proved because that the AUC under the ROC of the landslide susceptibility index calculated by CF-MLP model is 0.84;the study area is divided into three regions which are high sensitivity

medium sensitivity and low sensitivity on the basis of the forecast results of the model and it is consistent with the historical landslide data.Practice has proved that the CF-MLP model has solved the quantification problem of the landslide disaster pregnant environmental factors

and is beneficial to build a more accurate model for landslide susceptibility analysis by improving the convergence effect of the multiple layer perceptron networks.