矿井通风网络的反向增强型烟花算法优化研究

Research on opposition-based enhanced fireworks algorithm optimization for mine ventilation network

建立了以通风网络总能耗最小为目标的矿井通风网络非线性无约束优化模型。为提高该模型的优化能力和收敛速度,提出了一种反向增强型烟花算法。首先采用均匀反向初始化种群策略,将生成的均匀分布的随机种群和反向种群共同竞争,选择最优初始种群作为后续搜索的起始点;然后精细化控制烟花爆炸半径,使不同世代烟花种群的爆炸半径呈非线性递减,同代种群的爆炸半径由自身适应度值协调分配,并设定最小动态阈值以减少搜索资源浪费;最后采用精英反向学习选择策略,加强对精英烟花所在空间邻域的搜索,提高算法的全局勘测能力。实验结果表明,采用该算法对矿井通风网络进行优化后,在满足实际通风网络调节限制及用风需求基础上,总能耗可降低约23.2%,优化效果优于粒子群优化算法和增强型烟花算法。

A non-linear unrestraint optimization model of mine ventilation network was established which took the minimum total energy consumption of mine ventilation network as optimization objective.In order to improve optimization ability and convergence speed of the model,an opposition-based enhanced fireworks algorithm(OBEFWA) was proposed.Firstly,population initialization strategy based on opposition-based learning and uniform randomization is adopted,and uniform randomization population generated by the strategy is competed with opposition-based population,so that the optimal initial population is selected as starting point of subsequent search.Secondly,fireworks explosion radius is finely controlled,so that explosion radius of fireworks populations of different generations shows non-linear decline,and that of the same population generation is coordinated and distriblted according to their own fitness values.The minimum dynamic threshold is set to decrease waste of search resources.Finally,selection strategy of elite opposition-based learning is adopted to strengthen search for neighborhood of elite fireworks,so as to improve global exploration ability of the algorithm.The experimental results show that total energy consumption of mine ventilation network optimized by OBEFWA decreases about 23.2% which meets adjustment constraints and wind demand of actual ventilation network,and OBEFWA has better optimization effect than particle swarm optimization algorithm and enhanced fireworks algorithm.