针对不确定网络,研究具有随机参数的最短路径问题,采用随机数表示路径权值的不确定性,建立有约束的期望最短路模型.基于随机模拟方法,设计了一种融合退火技术的遗传算法,引入退火机制处理有约束的优化问题.在进化过程中,动态调节对不可行解的惩罚力度,使不可行解逐步被淘汰出去,最后收敛到问题的全局最优解.给出的数值实例验证了该算法的有效性.
Aiming at the uncertain networks, the shortest path problem with stochastic parameters is studied and the expected value model with constraints is constructed. Based on the method of stochastic simulation, a genetic algorithm integrating annealing method is presented to deal with the constrained optimization problem. The proposed algorithm introduces annealing mechanism to adjust dynamically the degree of the punishment to infeasible solutions, so that the infeasible solutions are eliminated gradually in the process of evolution and converges to a global optimal solution finally. The result of simulation test shows the algorithm is useful for solving practical problems.