小波網(wǎng)絡(luò)與多模型自適應(yīng)控制理論的研究.pdf

1、天津大學(xué)博士學(xué)位論文小波網(wǎng)絡(luò)與多模型自適應(yīng)控制理論的研究姓名：呂偉杰申請學(xué)位級別：博士專業(yè)：電力電子與電力傳動指導(dǎo)教師：劉魯源20021201AbStractThedevelopmentsofwaveletnetworksandmultiplemodelsarereviewedanalyzedandsummarizedBasedonwaveletnetworksandmultiplemodeladaptivecontroltheorie

2、s，theidentificationandcontrolmethodsforthecomplicatednonlineardynamicsystemsareproposedAaewlearningalgorithmofthecontinuewaveletnetworksparametersisproposed，thatis，hybridparameterslearningalgorithmTheparametersofwaveIetn

3、etworksaredividedintotwoparts，anddifferentmethodsareusedtotrainingthemComparetothetraditionalparameterslearningmethod，thenewmethodproposedinthispaperhastheadvantagesofrapidconvergencespeedandhighapproximationcapabilityTh

4、epropertiesofthewaveletnetworksareanalyzedAccordingtotheapproximationabilityofwaveletnetworks，thenonlinearstaticsystemandthenonlineardynamicsystemcanbeidentifiedTheparametersofthewaveletnetworksaretrainedbyusinghybridalg

5、orithmThewaveletnetworksareusedinidentifyingthestatorresistanceofinductionmotordirecttorquecontr01ThesimulationresultsshowthatthismethodiseffectiveandfeasibleThewaveletnetworksareusedascontrollertocontrolthenonlinearsyst

6、emThedirectcontrolandindirectcontrolmethodsarestudiedComparetothetraditionalneuralnetworks，thewaveletcontrolsystemshavetheadvantagesofhighcontrolprecisionandrapidconvergencerateThecontrollabilityandstabilityofwaveletnetw

7、orksareanalyzedandprovedTeducethemodelingerrorinthepiecewiselinearizationofnonlinearsystem，amultiplemodelswaveletnetworksidentificationmethodisproposedComparetothesinglewaveletnetworks，themultiplemodelswaveletnetworkscan