Application Of Neutral Network Based On Genetic Algorithm In Predicting Deformation Of Deep Foundation

In recent years, with the acceleration of the rate of expansion of urbanization, the floor space is scarce, more and more high buildings need to be built basements and underground parking and other facilities, so deep excavation has become inevitable for the study. Its stability is not only relate to the security of the upper structure, but also has an impact on the deformation of the surrounding buildings. Deformation of deep foundation is affected by many factors, such as envelope structure, construction conditions, temporal factors and environmental factors, etc., but it is difficult to determine the dominant factor of this deformation among certain factors. Therefore, the deformation of deep excavation is difficult to calculate with the unified empirical formula or mechanical model.This study uses neural networks excellent nonlinear mapping ability to solve the problem of deep foundation pit deformation prediction, this ability can process massively parallel untrained, noisy or incomplete data, which can make the neural network theory apply to deep foundation deformation prediction project. We build an artificial neural network model optimized by genetic algorithm to make up some defects of the neural network, and then apply it to deep deformation prediction. The combination of the two can not only play a generalization mapping ability of neural networks, but also make the neural network has a fast convergence and a strong learning ability, which is able to solve the complex nonlinear problem of deep deformation in theory.Meanwhile, based on Matlab R2010 b platform, the method which uses genetic algorithm to optimize neural network weights and thresholds, structures a located deep pit deformation prediction model(GA-BP model) using Matlab language. In order to verify the reliability of this model, we use a monitoring data observed in a deep foundation excavation process of a Beijing researching building engineering to predict the deformation of deep foundation, pointing at different types of support(CombinedSupport types and soil nailing wall type), and analysis the error through comparison,confirm the reliability of GA-BP model in deep excavation deformation prediction.