摘要
Thispaperdealswithon-linestateandparameterestimationofareasonablylargeclassofnonlinearcontinuous-timesystemsusingastep-by-stepslidingmodeobserverapproach.Themethodproposedcanalsobeusedforadaptationtoparametersthatvarywithtime.Theotherinterestingfeatureofthemethodisthatitiseasilyimplementableinreal-time.Theefficiencyofthistechniqueisdemonstratedviatheon-lineestimationoftheelectricalparametersandrotorfluxofaninductionmotor.Thisapplicationisbasedonthestandardmodeloftheinductionmotorexpressedinrotorcoordinateswiththestatorcurrentandvoltageaswellastherotorspeedassumedtobemeasurable.Real-timeimplementationresultsarethenreportedandtheabilityofthealgorithmtorapidlyestimatethemotorparametersisdemonstrated.Theseresultsshowtherobustnessofthisapproachwithrespecttomeasurementnoise,discretizationeffects,parameteruncertaintiesandmodelinginaccuracies.Comparisonsbetweentheresultsobtainedandthoseoftheclassicalrecursiveleastsquarealgorithmarealsopresented.Thereal-timeimplementationresultsshowthattheproposedalgorithmgivesbetterperformancethantherecursiveleastsquaremethodintermsoftheconvergencerateandtherobustnesswithrespecttomeasurementnoise.
出版日期
2005年02月12日(中国期刊网平台首次上网日期,不代表论文的发表时间)