| At present, CFG piles composite foundation is applied widely in the foundation reinforcement technology owing to its economizing material, being constructed conveniently and quickly, having good effect on the reinforcement and so on. But CFG piles compound foundation's design is an extremely complex processing. And the internal and foreign organizations have not yet carefully researched its theory, thus it is not formed as a unified design standard. So the design is carried on based on the project experience usually. In addition, the design of CFG piles compound foundation also is comprehensive strongly, contains the massive unascertained information, so design using the traditional method and mathematics plan method is very difficult to be effective. In such instances, the artificial nerve network is of use.The BP Neural Networks is used to the processing of CFG piles compound foundation's design in this paper. On the base of analyzing the factors affected CFG piles compound foundation's bearing capacity and settlement, a network model is put forward for CFG piles compound foundation'design. Because the established network model has so many inputs, the unascertained AHP is uses to decrease the number of the network model'inputs, for let the network model come true easily. Subsequently the actual collected projects are applied to train the nerve networks, and the obtained result is credible, thereby the non-linear relation between the factors which influent CFG piles compound foundation's bearing capacity and settlement is extracted effectively. Based on the nerve network model, this article analyzed the parameter's change rule affected the CFG pile compound foundation's bearing capacity and settlement is analysed, and the function between the parameter is gained using least squares method. Then the Simulated Annealing Alagorithmthe simulation is used to search the cost smallest design project in there rational scope. Thus a very effective method for CFG piles compound foundation's design is provided. |