Development Of A System Based Genetic Algorithm And Back Propagation Neural Networks For Vertical Bearing Capacity Prediction And Design Optimization Of Foundation Pile

With the development of high-rise buildings, large heavy industrial plants, large-span bridges and other projects, the application of pile foundation is more and more extensive. As an important indicator of testing the quality of pile foundation, the determination of vertical ultimate bearing capacity of foundation pile has also become the most important issue in the pile foundation design. As the mechanical properties and effect factors of vertical bearing capacity of rock-socketed piles are different from those of non rock-socketed, more attention has been paid on determining the pile vertical bearing capacity.In this paper, Genetic Algorithms are used to optimize the BP neural network, and the global search ability of genetic algorithm is combined with high-order non-linear property of BP neural network. This technique has been used to study the bearing capacity of foundation pile and a system has been developed for vertical bearing capacity prediction and design optimization of foundation pile.First, the effecting factors of vertical bearing capacity of foundation pile are selected to be as inputting parameters and the vertical bearing capacity of foundation pile as the outputting parameter. A highly nonlinear relationship between the inputing and outputing parameter is found by GA-BP neural network. As a result, two models, which proved precise for JiNan area, have been set up to predict the vertical ultimate bearing capacity of the rock-socketed and non rock-socketed foundation pile.Second, by using these two models, the effecting factors of vertical bearing capacity of non rock-socketed pile are analyzed.Third, based on the prediction model,further study has been done for the foundation pile optimization design.The sound size of a pile could be found which meeting either the requirement of maximal value of bearing capacity or the lowest cost under given design value of bearing capacity of foundation pile. Last, in the GUI, the interfaces for predicting and optimizing the vertical bearing capacity of foundation pile are set up, which used for rock-socketed or non rock-socketed piles. The script of this procedure is compiled to be as executable file, which could be run on the computer even without the MATLAB environment.