无迹西格玛点引导的拟反向黏菌算法及其工程应用

Unscented sigma point guided quasi-opposite slime mould algorithm and its application in engineering problem

针对黏菌算法搜索停滞和算法稳定性差等问题,提出了无迹西格玛点引导的拟反向黏菌算法。首先,使用了拟反向学习和拟反射学习两种反向学习过程,根据原始黏菌算法勘探和开采行为的表现时机,生成同时包含拟反向和拟反射的综合反向种群,扩大搜索范围;其次,根据种群的多样性程度判断是否使用反向种群重构原始种群进行后续计算,避免固定的反向过程破坏种群本身的搜索特点,提高搜索精度;最后,引入无迹变换的西格玛点,改进黏菌算法的基本移动模式,使无迹西格玛点引导黏菌算法的搜索,加快收敛速度。实验部分基于CEC2017基准测试函数,在传统统计特征和MAE排名、Wilcoxon秩和指标上验证算法的有效性;并在求解轿车侧面碰撞的实际工程优化问题上,与新颖的高水平群智能算法、改进算法、不完全算法进行对比测试。实验结果表明,改进策略有效且各策略间组合相得益彰,改进后算法的求解精度和鲁棒性更具竞争力。

Aiming at the search stagnation and poor stability of the slime mould algorithm,this paper proposed an unscented sigma point guided quasi-opposite slime mould algorithm.Firstly,it used quasi-opposite learning and quasi-reflected learning to exploration and exploitation behaviors according the original slime mould algorithm,to generate a comprehensive opposite population that included both quasi-opposite learning and quasi-reflected learning,and expanded the search space.Secondly,according to the diversity of the population,it decided whether to use the opposite population to regenerate a new population for subsequent calculations,avoided the continuous opposite process destroying the search characteristics of the population itself,and improved the search accuracy.Finally,it used unscented transformation sigma point to improve the basic movement mode of slime mould algorithm,made the unscented sigma point guide the search,and accelerated the convergence speed.The ex-perimental part used the CEC2017 benchmark test functions,it used traditional statistical index and MAE ranking,Wilcoxon rank-sum test to verify the effectiveness of the algorithm,and used it to solve the car side impact design problem,compared and tested with the novel swarm intelligence algorithms,improved algorithms and incomplete algorithm.The experimental results show that the strategies are effective and combinations of strategies complement each other,and the improved algorithm’s solution accuracy and robustness are more competitive.