Performance improvement of half bridge isolated dc dc converter using artificial intelligent techniques

Abstract

The switching power converter has profound impact on the power newlineconversion applications. Rising energy intensity leads to a higher cost for newlinedelivering power. Meanwhile, the demand for compact power supply is grown newlinesignificantly. It requires power supply with high efficiency, low profile and newlinehigh power density. Conventional PWM DC-DC converters have relatively newlinelow power density. In contrast, asymmetric half bridge converters have newlinenumerous advantages in DC-DC power conversions. In this work, half bridge isolated DC-DC converter operation, newlinemonitoring and its controller design is investigated to meet the challenges of newlinesoft switching, current sharing and high volumetric power density for newlineeffectively processing the energy. A converter topology which consists of two newlineasymmetric half bridge converters, whose output voltages overlap in a finite newlineinterval of time is proposed in this work to overcome the issues of newlineconventional circuits. This converter is robust to input voltage and operating newlineduty cycle variations. Both the power semiconductor switches of the proposed converter newlineoperate asymmetrically under zero voltage switching to achieve high newlineefficiency and low voltage stress. The ringing which resulted from the newlineoscillation between the transformer leakage inductance and the junction newlinecapacitance of two switches are eliminated. Also, the asymmetric structure of the proposed converter guarantees equal current sharing of the two rectifiers, newlinethereby maximizing the utilization of the output rectifiers. Conversion newlineefficiency is improved by providing a very small output filter and hence newlineelectrolytic capacitor of low reliability is not required for this circuit. Here, newlinethe main transformer is working as both fly back and forward transformer to newlinemake the transformer utilization high. This work investigates the effectiveness of an artificial neural network newlinewhile training as an asymmetric half bridge DC-DC converter circuit. The newlinenetwork is trained to form a mapping between the inputs and outputs of newlineconverter circuit using back propagation