Prediction Of Slope Stability Under The Conditions Of Rainfall Based On GA-LM-BP Neural Network

Rainfall is a main cause of slope instability.The rainfall infiltration will increase the water content of the slope,resulting in the increase of the soil weight and the decrease of shear strength parameters,thus affecting the stability of the slope.Because of the variability of soil weight,cohesion and angle of internal friction and the complexity of the slope instability mechanism under the condition of rainfall,some scholars have applied the BP neural network with strong nonlinear mapping ability to the prediction of slope stability and obtained some research achievements.However,because BP neural network has the drawback of slow convergence speed and weak generalization ability,so it is necessary to ameliorate BP neural network to improve the accuracy and efficiency of calculation.In this paper,the basic theory of BP neural network is introduced firstly,and the genetic algorithm and LM algorithm are proposed to improve the BP neural network.Then based on the procedures of the improved BP neural network algorithm,the corresponding calculation program is compiled,and the two slope example of program is discussed,the results showed that the program is reliable and the GA-LM-BP neural network convergence speed and generalization ability compared with the traditional BP neural network have been promoted Finally,with the background of Xiangqi hub project,the improved BP neural network is used to predict the stability of the slope.The main research contents and conclusions are as follows:(1)The basic theory of BP neural network and its improved algorithm are introduced.This paper introduces structure,algorithm and parameter selection of BP neural network,and puts forward genetic algorithm and LM algorithm to improve the network performance for the defects of BP neural network.(2)The improved BP neural network algorithm is implemented and verified.Firstly,the procedures of BP neural network and the improved BP neural network algorithm are introduced.Then the implementation of the above algorithm is programmed by MATLAB software.Finally,all kinds of procedures were examined based on two sample of slope example,the results show that the program is reliable and The BP neural network based on GA-LM algorithm is faster and more accurate than the traditional BP neural network and the improved BP neural network using only one of them.(3)Engineering Applications.The slope stability model of GA-LM-BP neural network was established by using the safety factor of the second-line ship lock of Xiangqi hub project as the target parameter,rainfall intensity,rainfall duration,slope angle and slope height as the change parameter.Then,the orthogonal method is used for program design,The SEEP/W and SLOPE/W modules in Geostudio are used to calculate the slope safety factor of soil weight and shear strength parameters changed under the conditions of rainfall.The results are used as sample of predicting slope stability model.Finally,after the model trained and tested,the model is applied to predict the safety factor of typical slope during the construction period,and the evolution rule of safety factor of every engineering slope with rainfall intensity and rainfall duration is obtained quickly,which is consistent with the actual change rule of safety factor under the rainfall.It is proved that the model has certain reliability and rationality.In the subsequent process of construction project in Xiang Qi,the prediction model can used on the slope safety factor for real-time prediction according to the excavation slope ratio,construction schedule,rainfall intensity and duration,which can obtained results quickly,the slope excavation protection measures during construction and the construction safety to provide technical support.It provides technical support for the use of protective measures and safety construction during the construction period of slope excavation.