Characteristic Parameters Analysis of Rockfall Movement for Rock Slope and LM-BPNN Forecasting Model

Abstract

Rockfall movement is influenced by many factors. Aiming at the rockfall movement characteristic parameters including horizontal movement distance, bounce height and motion energy, the orthogonal experimental method is introduced to study the sensitivity and unfavorable level combination of factors. Then the top five factors affecting the rockfall movement characteristic parameters are selected to establish the data set and the rockfall forecasting system for rock slope is established by Back-Propagation Neural Network, which can predict the characteristic parameters of rockfall movement. Finally, the forecasting system is verified by an engineering example. The results show that the orthogonal test method can determine the primary and secondary factors of rockfall movement characteristic parameters and the unfavorable level combination. Besides, forecasting system based on BPNN can easily estimate the parameter of rockfall movement without modeling and computing with Rockfall software.