|
- 2018
基于DE-RVM的开关磁阻电机无位置传感器研究DOI: 10.14081/j.cnki.hgdxb.2018.03.002 Keywords: 转子位置估计, 相关向量机, 差分进化算法, 无位置传感器, 开关磁阻电机estimating rotor position angle, relevance vector machine, differential evolution algorithm, sensorless, switchedreluctancemotor Abstract: 位置传感器的存在降低了开关磁阻电机可靠性与容错性,增加系统复杂度与成本. 针对这一问题,提出 了基于差分进化算法优化的相关向量机的开关磁阻电机转子位置估算方法. 采用相关向量机(relevancevector machine,缩写RVM)建立电机转子位置角预测模型,并利用差分进化(differentialevolution,缩写DE)算法 优化组合核函数参数,发挥其全局最优、快速收敛等优点,实现电机电流、磁链、转矩、转子位置角间的非 线性映射. 结果表明,该方法能够在无位置传感器情况下实现开关磁阻电机转子位置在线估算,且估算精度较 高、电机运行性能稳定.Thereliabilityandthefaulttoleranceofswitchedreluctancemotorareallreducedbytheexistenceofposi? tionsensors.Simultaneouslythecomplexityandthecostareincreased.Tosolvetheseproblems,themethodofbasedon relevancevectormachineoptimizedbydifferentialevolutionalgorithmispresentedfortheswitchedreluctancemotor. Thepredictivemodelofrotorpositionangleisbuiltbyrelevancevectormachine,andparametersofresultantkernelfunc? tionareoptimizedbyestimatingrotorpositionangledifferentialevolutionalgorithm,withadvantagessuchasglobalopti? mumandfastconvergence.Theproposedalgorithmcanachievenon-linearmappingofcurrent,flux,torque,androtoran? gularposition.Resultsshowthatthemethodcanachieveonlineestimateforrotorpositionanglewithoutpositionsensor. Andtheestimatedresultsarehighlyaccurateanditsrunningperformanceisstable.
|