AN INTELLIGENT ALGORITHM BASED ON COMPLEX SHAPE OPERATOR FOR OPTIMUM DESIGN OF SUPPORTING PILES

An algorithm for optimum design is developed through combining genetic algorithm with traditional complex shape search method. Neural networks are incorporated into the genetic algorithms and the computational procedures of genetic algorithms operators are adjusted. A practical deep excavation case is studied using the present algorithm. It is shown that the defects of genetic algorithms and artificial neural networks such as slow convergence and immersion into local solutions are done away with. Good global performance and satisfactory convergence rate are achieved.