Research on landslide hazard assessment based on PCA-MLP method

Landslide hazard assessment refers to the prediction of the degree of hazard posed by landslide disasters. It holds significant importance in landslide disaster prevention and control. A landslide risk assessment method based on PCA-MLP is proposed in this paper. Taking Ya'an City as the research area. Firstly, 12 landslide evaluation factors such as rainfall, geological lithology, and land type were selected. Secondly, correlation analysis was conducted to screen out landslide hazard evaluation index factors. Thirdly, principal component analysis (PCA) method was used to reduce the dimensionality of the original dataset to obtain the dimensionally reduced dataset. Finally, the original dataset and the dimensionally reduced dataset are separately input into evaluation methods such as decision tree (DT), logistic regression (LR), support vector machine (SVM) and three-layer neural network (MLP) for landslide hazard assessment studies. The accuracy and receiver operating characteristic (ROC) curve are used to compare the predictive accuracy of these evaluation methods. The results indicate that the evaluation method based on PCA-MLP has better evaluation accuracy. This method can provide a new approach for landslide hazard assessment.