摘要
本文设计了一种具有偏心磁极结构的永磁驱动器,采用三维有限元仿真建立改进的永磁驱动器的样本空间,通过中心复合试验设计选取合适的试验点,利用ANSYS有限元分析软件对试验点处的永磁驱动器重量、涡流损耗和输出转矩进行分析计算,获取响应值。根据响应值建立改进的永磁驱动器二阶响应面回归模型。并基于该回归模型,在确保永磁驱动器输出转矩不小于额定转矩的情况下,选择重量和涡流损耗功率最小为优化目标,利用自适应权重粒子群算法对永磁驱动器偏心磁极结构进行多目标优化,得出最优结构参数组合。有限元仿真实验对比改进前后永磁驱动器的磁通密度和涡流密度分布的结果,验证了该优化方法的可行性。优化结果使改进的永磁驱动器重量与原样机相比减少了1.145 kg,涡流损耗功率减小了70.76 W,提高了永磁驱动器的传动转矩,进一步提高了系统的工作效率。
A permanent magnet drive （PMD） with eccentric magnet pole structure was designed,3D finite element simulation was adopted to establish the sample space of the improved PMD,and then the Central Composite Design （CCD） experiment method was used to guide the selection of appropriate test points,The finite element analysis software ANSYS was utilized to analyze and calculate the weight,eddy current loss and output torque of the PMD at the test points,and obtain the response outputs of these samples; based on the response outputs the quadratic polynomials were employed to construct the quadratic response surface regression model.Based on this model,while ensuring that the output torque of the PMD is not less than the rating torque,the minimal eddy current loss and minimal weight were taken as the optimal objective; and then the adaptive weight particle swarm optimization （AWPSO） algorithm was used to perform multi-objective optimization of the eccentric magnetic pole structure of the PMD and obtain the optimal structure parameters of the PMD.In the finite element simulation experiment,the magnetic flux density and eddy current density distributions before and after the improvement were compared; the results prove that the parameter optimization method is feasible.Through the optimization,com pared with original prototype,the weight of the PMD decreases by 1.145 kg and the eddy current loss decreases by 70.76W.The optimization method has improved the drive torque and the working efficiency of the system.
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2014年第9期1963-1971,共9页
Chinese Journal of Scientific Instrument
基金
辽宁省技术创新重大项目(201309001)资助
关键词
永磁驱动器
有限元法
响应面法
自适应权重粒子群算法
多目标优化
permanent magnet drive
Finite Element Method （FEM）
response surface methodology （RSM）
adaptive weight particle swarm optimization （AWPSO）
multi-objective optimization
