中国电机工程学报
中國電機工程學報
중국전궤공정학보
ZHONGGUO DIANJI GONGCHENG XUEBAO
2014年
30期
5416-5424
,共9页
混合磁悬浮%电磁装置%永磁装置%粒子群%反向学习%多开端策略
混閤磁懸浮%電磁裝置%永磁裝置%粒子群%反嚮學習%多開耑策略
혼합자현부%전자장치%영자장치%입자군%반향학습%다개단책략
hybrid magnetic levitation%electromagnetic device%permanent magnetic device%particle swarm%opposition-based learning%multi-start strategy
混合磁悬浮装置的各项参数相互影响，决定着整个装置的性能。在满足承重要求的条件下，有必要对该装置的各项参数进行优化研究。为此，提出一种多策略改进粒子群算法，并将其应用到混合磁悬浮承重装置的参数优化中。首先，对混合磁悬浮装置进行介绍，通过分析永磁和电磁悬浮力，以励磁损耗和资金投入最小，和在允许范围内减载程度最高为目标，建立该装置的优化模型。在算法上，通过分析传统粒子群算法的缺陷，首次提出多开端策略来提高种群的多样性，结合反向学习和参数修正等多种策略对粒子群算法进行改进(多策略改进粒子群算法)，以广义Schwefel函数为验证函数，通过与其他粒子群算法的比较证明，改进算法具有更强的优势。最后，运用多策略改进粒子群算法对磁悬浮模型进行优化，将优化结果与原有参数进行比较，分析可知该结果更加符合实际情况，通过仿真验证该结果的合理性，为进一步建立实验模型奠定了理论基础。
混閤磁懸浮裝置的各項參數相互影響，決定著整箇裝置的性能。在滿足承重要求的條件下，有必要對該裝置的各項參數進行優化研究。為此，提齣一種多策略改進粒子群算法，併將其應用到混閤磁懸浮承重裝置的參數優化中。首先，對混閤磁懸浮裝置進行介紹，通過分析永磁和電磁懸浮力，以勵磁損耗和資金投入最小，和在允許範圍內減載程度最高為目標，建立該裝置的優化模型。在算法上，通過分析傳統粒子群算法的缺陷，首次提齣多開耑策略來提高種群的多樣性，結閤反嚮學習和參數脩正等多種策略對粒子群算法進行改進(多策略改進粒子群算法)，以廣義Schwefel函數為驗證函數，通過與其他粒子群算法的比較證明，改進算法具有更彊的優勢。最後，運用多策略改進粒子群算法對磁懸浮模型進行優化，將優化結果與原有參數進行比較，分析可知該結果更加符閤實際情況，通過倣真驗證該結果的閤理性，為進一步建立實驗模型奠定瞭理論基礎。
혼합자현부장치적각항삼수상호영향，결정착정개장치적성능。재만족승중요구적조건하，유필요대해장치적각항삼수진행우화연구。위차，제출일충다책략개진입자군산법，병장기응용도혼합자현부승중장치적삼수우화중。수선，대혼합자현부장치진행개소，통과분석영자화전자현부력，이려자손모화자금투입최소，화재윤허범위내감재정도최고위목표，건립해장치적우화모형。재산법상，통과분석전통입자군산법적결함，수차제출다개단책략래제고충군적다양성，결합반향학습화삼수수정등다충책략대입자군산법진행개진(다책략개진입자군산법)，이엄의Schwefel함수위험증함수，통과여기타입자군산법적비교증명，개진산법구유경강적우세。최후，운용다책략개진입자군산법대자현부모형진행우화，장우화결과여원유삼수진행비교，분석가지해결과경가부합실제정황，통과방진험증해결과적합이성，위진일보건립실험모형전정료이론기출。
The mutual influences among the parameters of hybrid magnetic levitation devices determine the device performances. When the load-reduction requirement is met, it is necessary to optimize structural parameters of the devices. A novel improved multi-strategy particle swarm algorithm is proposed, and is applied on the structural optimization of devices. Firstly, the principle of devices is introduced, the electromagnetic force and permanent magnetic force are analyzed. Based on the idea of minimal electromagnetic loss and investment minimum, and maximal load-reduction degree within the scope allowed, optimization objective functions of the devices are built. Then, by analyzing the defects of the traditional particle swarm algorithm, this paper proposes a multi-start strategy to improve the diversity of the swarm. The technology combines with opposition-based learning and parameter correction to improve the particle swarm algorithm (i.e. improved multi-strategy particle swarm algorithm). To verify the rationality of the improved strategies and its effectiveness of the improved algorithm, the generalized Schwefel function is taken as the test function, solutions are compared with other particle swarm algorithms, the result is proven that the improved algorithm has more superiority. Finally the improved multi-strategy particle swarm algorithm is applied to optimize the model. Through compared the optimization results with original parameters, it is concluded that the results are more practical on space utilization, which is verified by FEM. The results lay theoretical basis for further establishment of experimental models.
