基于非支配排序的蜣螂优化算法  

Mantis optimization algorithm based on non-dominated sorting

作　　者：丘斯帆 QIU Sifan(School of Electrical and Control Engineering,Heilongjiang University of Science and Technology,Harbin 150022,China)

机构地区：[1]黑龙江科技大学电气与控制工程学院,哈尔滨150022

出　　处：《黑龙江电力》2024年第5期421-427,共7页Heilongjiang Electric Power

摘　　要：现实社会中的问题往往不止单一目标而是复杂的多目标问题,解决这类问题就需要一种高效的优化算法。为此,基于蜣螂优化算法,提出一种非支配排序的蜣螂优化算法(NSDBO),以解决多目标优化问题。结合NSGA-Ⅱ算法中的非支配排序方法,对蜣螂种群进行非支配排序;将传统外部存档策略替换为动态外部存档,并提出一种新的拥挤度距离公式,增强算法全局寻优能力,同时维持种群的多样性。为了验证NSDBO算法的效率和有效性,选取ZDT系列及DTLZ系列测试函数对其进行测试,结果表明其收敛精度提升了21.5%~44.3%。将其应用于变压器的故障诊断,其结果表明ND-SVM相比于SVM有更高的准确率,符合工程应用的标准。Problems in the real society are often not only a single goal but a complex multi-objective problem.Solving such problems requires an efficient optimization algorithm.Based on the mantis optimization algorithm,a new mantis optimization algorithm(NSDBO)is proposed to solve the multi-objective optimization problem.Based on the non-dominant ranking method in NSGA-II algorithm,the traditional external archiving strategy is replaced with dynamic external archiving,and a new crowding distance formula is proposed to enhance the global optimization ability of the algorithm and maintain the diversity of the population.In order to verify the efficiency and effectiveness of NSDBO algorithm,ZDT series and DTLZ series test functions were selected to test it,and the results show that the convergence accuracy is improved by 21.5%~44.3%.It is applied to the fault diagnosis of transformer,and the results show that ND-SVM has higher accuracy than SVM and meets the standard of engineering application.

关 键 词：蜣螂算法 非支配排序 动态外部存档 拥挤度距离 故障诊断 

分 类 号：TM411[电气工程—电器] TP18[自动化与计算机技术—控制理论与控制工程]