My Research During the Postgraduate and the Master's Thesis

The current rapid development of the domestic economy sustained and stable，various types of projects such as railway tunnel, highway and large-scale water conservancy and hydropower projects have also started construction in the country，and the safety problem of slope engineering becom very important. How to ensure the safety and stability of the slope engineering technicians and researchers should focus on the problem. In the current methods of slope stability analysis, Artificial intelligence is an effective and accurate way. In this paper, the stability analysis of slope based on neural network and genetic algorithm is studied. Principal component analysis (PCA) is mainly based on linear transformation, The number of indicators of the data space to a small number of principal components. Information about the data represented by these principal components. To achieve the purpose of reducing the correlation between the data and redundant information, reducing the input of the genetic neural network.This can reduce the complexity of the problem to a certain extent.At the same time, the prediction accuracy of GABP neural network is improved.Genetic neural network (GABP) by using the genetic algorithm to the traditional BP neural network optimization, which can complex system self-organization and adaptive function to deal with all kinds of nonlinear information reference neural network at the same time, with the performance of genetic algorithms global search is an effective method to make up the BP neural network training speed is slow, sensitive and easy to fall into the defect the local minimum value of the initial value. In this paper, the research methods, current situation and existing problems of slope stability are introduced. The main research contents and research route.The basic concepts and principles of neural network and genetic algorithm are introduced, and the basic principle and characteristics of principal component analysis (PCA) are introduced.The training samples are processed by PCA, and the 6 input vectors are reduced to 2. After the combination of genetic algorithm to optimize BP neural network structure, genetic neural network (GABP), and the principal component analysis of data import processing after the training, with the accuracy of test samples to validate the model to forecast the slope safety factor. Secondly, using the genetic neural network without principal component analysis and the radial basis function neural network to predict the slope safety factor, the author makes a comparative analysis of the three factors from the angle of error. Then the AIC information criterion method is introduced to analyze the prediction model of the three algorithms.Results show: Using the model of PCA-GABP neural network performance better, the prediction results are better than the radial basis function (RBF) neural network, genetic neural network (GABP) has higher prediction accuracy. Finally, slope engineering in Shijingshan District Shimen road Beijing city section, combined with the actual engineering situation, extract the analysis model of the required data and import the model, using numerical simulation method FLAC3D do comparative analysis, the results showed that the prediction model of the data is more accurate and feasible for application in the engineering practice.

Slope Stability; Principal Component Analysis; Neural Network; Genetic Algorithm; AIC.