Study On Wavelet Networks Applied In Buck Converters

The discourse firstly analyzed the principle of the soft-switching Buck converterand introduced the actuality of the controlling of the systems of non-linear anduncertain factors of the soft-switching power converters, used module theory toanalyse the operation state of voltage-feed ZCS Buck converter, reduced the difficultyof constituting the vary state equation since influent of resonance inductance andcapacitance. The time-discrete mathematic modeling of ZCS Buck converter has beenset up by the way of L-transformation. The power converters' working conditionunder the state of soft-switching has been dissertated, and simulation experimentationwas carried out. The result and the difference of the steady state and unsteady statewere analyzed through the wavelet tools, in order to make better use of the waveletnetworks control.Since the wavelet networks consists of the learning and adaptive capability ofneural nets and the frequency character and zoom ability of the wavelet analyse, itwas provided with better the approach performance and rate. In theory the paperexpatiated on the frame, the function choice, the initialization of the parameter, thelayer nodes and the arithmetic of the parameters, the wavelet networks were optimaldesigned. The theoretic study and experimentation were gone along with the waveletnets control Buck converter and the two models of control were constituted, one was asimple Buck converter that applied the wavelet networks control, the other was theBuck ZCS-PWM converter that was controlled by the fuzzy wavelet networks. Thewavelet networks control applied the learning algorithm that is made ofLevenberg-Marquart (LM) algorithm and least square method;the fuzzy waveletnetworks control wielded two segments study strategy and adopted extended Kalmanfilter (EKF) algorithm to adjust nonlinear parameters and least square method tomodify all network weights. Simulation experiments was carried out under the precisemathematics models, and the results was compared with the voltage-feed control'sconverter.The paper analyzed the soft-switching converter traits and the wavelet networkscontrol theory, introduced the wavelet theory into the DC/DC converters and thedifference of the steady state and unsteady state of Buck converter working wasanalyzed. The wavelet networks was applied on the Buck converters, and the controlcircuit models were designed successfully and simulated, optimized the controlparameters and achieved the intelligentized control of the power supply system.