Landslide Forecasting Model Based on PCA and Improved CS-RBF

Abstract

Landslide disasters pose a serious threat to human life and property, and strengthening effective forecasting of landslide disasters is of great significance. Taking the landslide monitoring points in Shanyang County, Shaanxi Province as an example, this study proposes a landslide probability forecasting model based on principal component analysis (PCA) and cuckoo search (CS) optimized radial basis function (RBF) neural network. Firstly, the main influencing factors of landslide disasters in the area are determined, and the PCA algorithm is used to reduce the dimensionality of landslide influencing factors to avoid the problem of model redundancy caused by excessively large data dimensions. The dimensionality-reduced data is then input into the RBF neural network for landslide probability forecasting. Secondly, an improved Cuckoo Search algorithm is used for parameter optimization to improve the accuracy of landslide probability forecasting. Various models including back propagation (BP), RBF, genetic algorithm-RBF (GA-RBF), CS-RBF, and others are compared with the improved CS-RBF model through experimental analysis. The results show that the predictive performance of the CS-RBF model is superior to the other models, with a root mean square error of 0.017 56 and an average absolute error of 0.011 78. This model exhibits higher reliability, providing strong support and guarantee for the practical application of landslide early warning.

Keywords

• landslide forecasting;PCA algorithm;RBF neural network;improved CS-RBF