面向混合变量的蚁群优化方法及其在武器装备体系规划中的应用

The solution of complex operational problems including System of Armament Systems Planning urgently require optimization methods of high performance, which can sufficiently deal with mixed continuous and ordinal or categorical discrete variables (mixed variables, for short). On the basis of the Ant Colony Optimization Metaheuristic, this project investigates the mixed variables oriented Ant Colony Optimization, breaking through the limitation of the original method only for optimization of discrete variables. Scientific problems such as pheromone modeling, pheromone positive-feedback updating mechanisms, ant probabilistic solution construction approaches for mixed variables will be solved. Key technical issues on constructing the set of mixed-variable benchmark functions and designing the indicator of anytime optimization performance evaluation for algorithm parameter configurations will be settled. Combining with the Iterated F-Race method, this project studies automatic algorithm parameter configuration for the new Ant Colony Optimization algorithm. Using the new set of mixed-variable benchmark functions and the mixed-variable benchmark engineering optimization problems, this project conducts empirical studies on the performance evaluation of the new Ant Colony Optimization method. Basing on the quantitative capability requirement modeling, optimization and evaluation from System of Armament Systems Planning, this project explores the typical application of the mixed variables oriented Ant Colony Optimization, demonstrating its effectiveness and feasibility. This project will not only expand applicability of the Ant Colony Optimization theory, but also provide new optimization method and tool of high performance for solving complex mixed-variable problems from System of Armament Systems Planning and other realistic applications, which is of great theoretical significance and important practical value.

武器装备体系规划等复杂运筹问题的解决迫切需要充分处理混合连续、有序或无序离散变量（简称混合变量）的高性能优化方法。本项目基于蚁群优化元启发式理论，突破其仅对离散变量进行优化的局限，研究面向混合变量的蚁群优化方法。解决面向混合变量的信息素模型构建及其正反馈更新机制、面向混合变量的蚂蚁随机解构建方法等科学问题；突破混合变量标准函数集构建、面向任意时间优化评估的参数性能指标设计等关键技术；结合“迭代式弗里德曼竞赛”方法，进行新的蚁群优化算法参数自动化配置研究；运用新的混合变量标准函数集和标准工程优化问题，进行新的蚁群优化方法性能实证研究；结合武器装备体系规划中的定量能力需求建模和优化评估问题，开展新的蚁群优化方法典型应用研究，验证其有效性和可行性。本项目将拓展蚁群优化理论的适用性，为解决武器装备体系规划等现实应用中的复杂混合变量优化问题提供高性能的优化方法和工具，具有重要理论意义和应用价值。

武器装备体系规划等复杂运筹问题的解决迫切需要充分处理混合连续、有序或无序离散变量（简称混合变量）的高性能优化方法。本项目基于蚁群优化元启发式理论，研究面向混合变量的蚁群优化方法，突破了其仅对离散变量进行优化的局限，解决了面向混合变量的信息素模型构建及其正反馈更新机制、面向混合变量的蚂蚁随机解构建方法等科学问题；突破了混合变量标准函数集构建、面向任意时间优化评估的参数性能指标设计等关键技术；实现了新蚁群优化算法基于“迭代式弗里德曼竞赛”方法和任意时间优化评估的参数自动化配置；设计了新的混合变量标准函数集，并运用了新的混合变量标准函数集和标准工程优化问题，进行新的蚁群优化方法性能实证研究；结合武器装备体系规划问题，开展了新的蚁群优化方法典型应用研究。本项目拓展了蚁群优化理论的适用性，为解决武器装备体系规划等现实应用中的复杂混合变量优化问题提供高性能的优化方法和工具，具有重要理论意义和应用价值。

内容获取失败，请点击重试