改进粒子群算法目标函数变化分类动态优化.docVIP

改进粒子群算法目标函数变化分类动态优化 　　摘 要： 由于优化问题的目标函数和约束条件都随着时间而改变导致其最优值也发生改变，提出一种基于改进粒子群算法的目标函数变化分类动态优化算法。首先对动态优化问题进行定义，明确问题的研究对象，提出对目标函数随时间变化程度分类的思想，通过对变化的函数进行监测的方法将其分为剧烈变化、中等程度变化和弱变化三种类型，并针对不同的强度变化对粒子群算法采用不同的改进策略，最后将不同的策略融入计算。通过采用移动多峰问题进行测试，结果表明，提出的改进粒子群优化算法能监测目标函数变化，并能随时跟踪到最优解，平均离线误差相对于标准粒子群算法更小，性能更稳定 关键词： 粒子群算法； 动态优化； 目标函数时变分类； 移动峰问题 中图分类号： TN911.1?34； TP301.6 文献标识码： A 文章编号： 1004?373X（2017）07?0175?04 Dynamic optimization of objective function changing classification based on improved particle swarm optimization SU Yu， KONG Guoli （College of Information Engineering， Zhongzhou University， Zhengzhou 450001， China） Abstract： The objective function and constraint condition for the optimization problem are changed with time， and may change its optimal value. A dynamic optimization of the objective function changing classification based on improved particle swarm optimization is proposed. The dynamic optimization problem is defined to determine the study object of the problem. The classification thought that the objective function is changed with the time varying degree is put forward. The varying function is divided into the types of drastic change， medium grade change and weak change with the monitoring method. Different strategies are adopted for the particle swarm optimization according to the different intensity changes， and integrated for computation. The algorithm was tested with the moving multi?peak problem. The test results show that the improved particle swarm optimization can monitor the changes of the objective function， track the optimal solution momentarily， its average offline error is smaller than that of the standard particle swarm optimization algorithm， and the performance is more stable. Keywords： particle swarm optimization； dynamic optimization； time varying classification of objective function； moving peak problem 0 引 言 由于在很多实际生产过程中遇到的优化问题都在不停变化，同时其目标函数和约束条件也会随着时间在不停的发生着变化，最终导致其问题的最优解改变[1?3]。如动态旅行商问题（DTSP）[4]中旅行城市的增加或减少；车辆路径规划问题（DVRP