A novel excitation controller using support vector machines and approximate models-使用支持向量机和近似模型的新激发控制器.pdfVIP

JControl 2008 TheoryAppl DOI10．1007／sl1768—008．7064．4 Anovelexcitation controller vector usingsupport machinesand models approximate LI YUAN，YaonanWANG，Shutao Xiaofang ofElectricalandInformation Hunan (CoUege Engineering，HunanUniversity,Changsha410082，China) a Abstract：This novelexcitationcontroller vector and paperproposes usingsupportmachines(SVMlapproximate models．Thenonlinearcofltrollawisderived basedan on metllodvia directly input—outputapproximationTaylorexpansion． whichnot avoids contr01 andintensive alsoavoidsonline or only complexdevelopment computation．but learningadjust． a SVM isinvolvedinbothmodelidentificafionandcontroller ment．Onlygeneralmodellingtechnique implementation． Therobustnessofthe is establishedtIle method．SeveralsimulationsdemonstratetIle stabilityrigorously usingLyapunov of effectivenessthe excitationcontroller． proposed vector machines；Nonlinear Keywords：Support control；Approximatemodel；Neuralnetworks；Identification 1 Introduc