若干优化算法的运行分析比较

Comparison of Operational Behavior for Several Optimization Algorithms

研究了若干优化算法在连续空间寻优问题中的应用。通过对标准寻优问题的寻优结果,分析并比较了优化算法的运行效果。考察的性能指标主要有寻优偏差、寻优次数等,分析比较方法采用平均分析法和最优最差分析方法等。文中涉及到的优化算法有遗传算法、模拟退火法、禁忌搜索法、单纯形法和蚁群算法。实验结果表明,单纯形法受解空间的维数以及寻优问题类型的影响较大,而遗传算法、模拟退火法、禁忌搜索法和蚁群算法对各个标准寻优问题则各有所长。对于工程中的实际寻优模型,可将其与标准寻优问题相对照以选择较适合的算法。

The application of several optimization algorithms in continuous function optimization is studied. Based on the operational results for some standard optimization problems, the effects of the optimization algorithms are compared. The guide lines considered are dispersion of objective function value and searching times, and the average analytical method and the worst - best analytical method are used. Genetic Algorithm, Simulated Annealing, Tabu Search, Simplex Mgorithm, Ant System Algorithm are included. Experiment results show that Simplex Algorithm is more sensitive to the dimension of the problems, while the others are fit for certain models, comparatively. As to the natural optimization models in engineering, it is suggested to choose suited algorithms comparing to the standard problems.