Modeling and Control of Series Resonant Converter for High Voltage Application Using Adaptive Sliding Mode Controller

Abstract

Inductor inductor capacitor (LLC) series resonant converter is used for high frequency and high voltage applications. The converter operates in zero voltage swithching (ZVS) region to reduce swithching loss happened in swithches used to turn ON and OFF the system. This operation makes the output voltage to be regulated for wide input voltage and load resistance variation with small variation of switching frequency. This is the main advantage of LLC resonant converter compared to other resonant converters. But, LLC series resonant DC-DC converter have a problem of the voltage regulation due to nonlinearity, the uncertainity and disturbance which is caused by the deviation of input voltage (300-400)V, load resistance (3-8)kΩ and resonant components from the nominal value. Extended describing function and Jacobian Linearization methods is used to develop small signal modeling of the converter. But, the linearized system has seven states wich are not measurable and needs obsrever for controller design. Hence, seven state small signal modeling of the converter is reduced to two measurable states using the relationship between the state variables. Model reference adaptive sliding mode controller is designed for simplified second order system of LLC converter to control the output voltage and rectifier current. The lyapunov adaptive method is used to estimate the sliding mode controller parameters to solve the disturbance caused by parameter variation in the converter. The stabilty of controller is checked using Lyapunov stability criteria. The converter is designed for 4kV output voltage, 0.5A output current and 2kW output power. The simulation is done using MATLAB/SIMULINK® software. The simulation result indicates that the adaptive sliding mode controller has better performance compared to supper twiting sliding mode controller. The setling (ts) is reduced to 2.5msec and rise time (tr) is increased to 334.24 sec respectively, while the steady state error (ess) is reduced to 0.2%.