基于复合形算子的基础支护桩优化设计智能算法研究

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

本文通过遗传算法和传统复合形搜索法相结合,基于对遗传算法算子计算结构的调整,并将遗传算法与神经网络相结合,提出并研究了一种新的优化设计方法,协同求解复杂工程中的优化问题。并针对悬臂式支护桩的优化设计的数学模型,采用该算法进行了优化设计分析;计算结果表明,该算法可克服遗传算法最终进化至最优解较慢和人工神经网络易陷入局部解的缺陷,具有较好的全局性和收敛速度。

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.