Prediction of Landslide Risk Based on Modified Generalized Regression Neural Network Algorithm

Abstract

As a natural disaster, landslide causes immeasurable losses. The particle swarm optimization algorithm (PSO) and genetic algorithm (GA) are used to modify the smoothing factor of the generalized regression neural network (GRNN), which improves the prediction efficiency of GRNN. By building a landslide monitoring platform, the rainfall, shallow soil moisture content, deep soil moisture content, soil glide stress, and surface displacement are used as five landslide factors for landslide risk analysis, and the modified landslide models are applied to the processing of landslide data. These two models are used to predict the landslide risk, and compared with the models of BP, Elman neural network and RBF neural network for landslide prediction. The results illustrate that the modified GRNN landslide models have better prediction effects of landslide risk than BP neural network model, Elman neural network model, and Radial basis function neural network model, which provides a reference for engineering practice.