Thispaperdealswithon-linestateandparameterestimationofareasonablylargeclassofnonlinearcontinuous-timesystemsusingastep-by-stepslidingmodeobserverapproach.Themethodproposedcanalsobeusedforadaptationtoparametersthatvarywithtime.Theotherinterestingfeatureofthemethodisthatitiseasilyimplementableinreal-time.Theefficiencyofthistechniqueisdemonstratedviatheon-lineestimationoftheelectricalparametersandrotorfluxofaninductionmotor.Thisapplicationisbasedonthestandardmodeloftheinductionmotorexpressedinrotorcoordinateswiththestatorcurrentandvoltageaswellastherotorspeedassumedtobemeasurable.Real-timeimplementationresultsarethenreportedandtheabilityofthealgorithmtorapidlyestimatethemotorparametersisdemonstrated.Theseresultsshowtherobustnessofthisapproachwithrespecttomeasurementnoise,discretizationeffects,parameteruncertaintiesandmodelinginaccuracies.Comparisonsbetweentheresultsobtainedandthoseoftheclassicalrecursiveleastsquarealgorithmarealsopresented.Thereal-timeimplementationresultsshowthattheproposedalgorithmgivesbetterperformancethantherecursiveleastsquaremethodintermsoftheconvergencerateandtherobustnesswithrespecttomeasurementnoise.
