面向混合变量和任意时间优化的蚁群算法

Mixed variable and any time optimization oriented ant colony optimization algorithm

针对现实问题中优化模型复杂、变量类型混合、求解难问题,通过构建面向混合变量的蚁群优化信息素模型和设计蚂蚁随机解构建方法,提出能够充分有效处理混合连续、有序或无序离散变量的蚁群优化算法。进一步考虑现实问题中目标函数评估次数未知或昂贵优化场景,设计面向任意时间优化的算法参数评估指标,自动化配置算法同时提高解的质量和优化执行效率,生成了面向混合变量和任意时间优化的蚁群算法。最后在标准工程优化问题中进行测试,通过与文献结果的比较,验证了新蚁群算法的高效性和鲁棒性。

Aiming at solving difficulties from the complexity of optimization model and the mixture of variable types in the practical problems, an ant colony optimization algorithm is proposed that can sufficiently deal with mixed continuous and ordinal or categorical discrete by constructing the pheromone model and designing ant probabilistic solution construction approaches for mixed variables. Then a mixed variable and any time oriented ant colony optimization algorithm is generated by further considering the unknown number of objective function evolutions and expensive scenario in the practical problems, the any time optimization oriented evaluation indicator is designed for algorithm parameters and automatically configure algorithm for improving solution quality and optimization efficiency. Finally, the generated algorithm is tested on various engineering optimization benchmark problems. Compared with results from the literature, the proposed algorithm demonstrates its effectiveness and robustness.