Genetic Algorithm-Based Research Of Prediction On Vertical Ultimate Bearing Capacity Of PHC Pipe Piles

PHC pile, having great development space, is widely used in China with a series of distinct advantages, such as high bearing capacity, wide range of application, low cost, good durability, and so on. Due to the test cost, time, experimental conditions, as well as the requirements of continuing to bear the building structure, PHC piles are often not pressed to destroy, without vertical ultimate bearing capacity. Special prediction research on PHC pipe pile bearing capacity, especially with interdisciplinary knowledge is still in the initial stage of discussing.Genetic algorithm is an adaptive probability optimization method of a simulation of a natural evolutionary process to search optimal solution. It has the advantage of intrinsic implicit parallelism, not relying on gradient and continuity of function, fast searching ability and global optimization capability. It is necessary to be introduced into the practical engineering issues for further research and improvement.The development background, characteristics, basic principle and operation process of genetic algorithm are introduced and the basic genetic algorithm is tested through an optimization test example in this paper. Basic genetic algorithm has the weak optimization ability and premature problems.The improvement strategy, changing the encoding method into Gray code, combination of the elitist strategy and optimal reservation strategy, alternately selecting with two kinds of selection methods and the adaptive crossover and mutation probability, is put forward in this paper, and then matlab language is used to complete the improved genetic algorithm programming. The program, combined with the specific project example, is used to optimize three curve prediction models of PHC pile vertical ultimate bearing capacity, the value of which is calculated according to "technical code for testing of building foundation piles". And finite element method is used to simulate the two groups, both of which will be compared with results of the general programming solver. The examples show that it is feasible to solve the curve prediction models of PHC pile vertical ultimate bearing capacity with improved genetic algorithm program, and the results of Usher curve prediction model which tend to safety are effective and reliable.