The design of Class-EF₂ inverters using multi-objective optimization

This thesis explores the use of multi-objective optimization algorithms for the design of high frequency inverters. A state-space model of the ideal Class-EF₂ inverter is derived and its accuracy is validated by MATLAB and LTSpice simulation. The model is then applied to the Multi-Objective Genetic Optimization (MOGO) and Multi-Objective Particle Swarm Optimization (MOPSO) algorithms to design three inverters with varying output power, frequency, and load requirements. The final designs are compared with analytical results to verify the optimization-based design approach. The ideal state-space model is then extended to include the parasitic elements of components, and further extended to consider the internal resistances and capacitances of the switch. These new models are applied to the MOGO and MOPSO algorithms to design the same three inverters as the ideal case. The final designs are simulated in LTSpice to evaluate their performance, and comparisons are presented to demonstrate the effects of the parasitic elements and switching dynamics on the component values and overall circuit operation. A design example is also presented to demonstrate the design of a 6.78 MHz, 100W, 20 Ω Class-EF₂ inverter, and provide designers with insight on how to apply the proposed design approach to their own designs