施工导流建筑物优化设计是一个复杂的非线性约束优化问题．针对传统优化方法的局限性，提出了施工导流建筑物优化设计的退火遗传算法，即改进遗传算法；保存每代的最优个体，并且引入模拟退火技术，不仅确保了解的最优性，而且克服了一般二次型罚函数收敛性能不好的缺点．实例验证了此法应用于施工导流建筑物优化设计的可行性，同时通过与一般二次型罚函数遗传算法的比较，说明本方法具有更好的收敛性能． The construction diversion design optimization is a complex nonlinear optimization problem. Aiming at the limitations of the traditional optimization methods, an annealing genetic algorithm is proposed to optimize the design of construction diversion buildings, that is, to improve the genetic algorithm. To preserve the best individual of each generation and to introduce simulated annealing techniques, not only the optimality of understanding, Overcomes the shortcomings of poor convergence performance of general quadratic penalty function. An example is given to demonstrate the feasibility of this method in the optimization design of diversion structures. The comparison with the general quadratic penalty genetic algorithm shows that the proposed method has better convergence performance.