2023年第5期仪表技术与传感器InstrumentTechniqueandSensor2023No.5基金项目:宁夏自然科学基金项目(2021AAC03061,2022AAC03118);国家自然科学基金项目(61863030)收稿日期:2022-11-04基于改进麻雀搜索算法的粒子滤波研究石硬,韩玉兰,薛丽(宁夏大学物理与电子电气工程学院,宁夏银川750021)摘要:针对标准粒子滤波存在的粒子贫化现象提出了一种基于改进麻雀搜索算法的粒子滤波算法(SSA-PF)。该算法将粒子状态值看作麻雀个体位置,使粒子滤波的状态估计转变成麻雀种群的觅食寻优。首先,使用Logistic-tentmap初始化分布;其次,加入比例系数解决当最大迭代次数较小时发现者收敛速度过快的问题,同时受高斯变异与自适应权重的启发提出一种新的发现者更新策略;然后,改进追随者更新公式,使种群分布更契合粒子滤波的求解思想;最后,引入随机游走策略有效帮助算法跳出局部最优。仿真结果证明:SSA-PF的均方根误差更小,粒子的分布更合理,并且减少了状态估计所需样本数量。关键词:粒子滤波;粒子贫化;麻雀搜索算法;重采样;自适应权重;状态估计中图分类号:TP18文献标识码:A文章编号:1002-1841(2023)05-0076-08ResearchonParticleFilterBasedonImprovedSparrowSearchAlgorithmSHIYing,HANYu-lan,XUELi(SchoolofPhysicsandElectronic-ElectricalEngineering,NingxiaUniversity,Yinchuan750021,China)Abstract:Aparticlefilteringalgorithm(SSA-PF)basedontheimprovedsparrowsearchalgorithmwasproposedtosolvetheparticledilutioninthestandardparticlefiltering.Inthisalgorithm,theparticlestatevaluewasregardedastheindividualposi-tionofsparrow,andthestateestimationofparticlefilterwastransformedintotheoptimizationofsparrowpopulationforaging.First,thedistributionwasinitializedusingLogistic-tentmap.Secondly,aproportionalcoefficientwasaddedtosolvetheproblemthatthediscovererconvergencerateistoofastwhenthemaximumnumberofiterationsissmall,andanewdiscovererupdatestrategywasproposedinspiredbyGaussianvariationandadaptiveweight.Then,thefollowerupdateformulawasimprovedtomakethepopulationdistributionmoreconsistentwiththesolutionideaofparticlefilter.Finally,therandomwalkstrategywasintro-ducedtohelpthealgorithmescapefromthelocaloptimal.Simulationresultsshowthattherootmeansqua...