A Comprehensive Study of Dual Active Bridge Converter and Deep Belief Network Controller for Bi-directional Solid State Transformers

Abstract

This dissertation presents a comprehensive study of Dual Active Bridge (DAB) converter and Deep Belief Network (DBN) controller for bi-directional Solid State Transformers (SSTs). The first contribution is to propose a dc-dc DAB converter as a single stage SST. The proposed converter topology consists of two active H-bridges and one high-frequency transformer. Output voltage can be regulated when input voltage changes by phase shift modulation. Power is transferred from the first bridge to the second bridge. It analyzes the steady-state operation. The second contribution is to develop an average model for dc-dc DAB converters. The transformer current in DAB converter is purely ac, making continuous-time modeling is difficult. Instead, the proposed approach uses the only 1st order terms of transformer current and capacitor voltage as state variables. The third contribution is the controller design of a dc-dc DAB converter. The PI gains are allowed to vary within a predetermined range and therefore eliminate the problems from the conventional PI controller. The performance of the proposed artificial intelligence gain scheduled PI controller is simulated and compared with the conventional fixed PI controller under steady state error, responding time and load disturbances. The experimental system of DAB converter is implemented using digital signal processing unit, Texas Instrument TMS320F28335 control board, to examine and verify the performance of the proposed controller under various operating conditions. Simulation and experimental results show a good improvement in transient as well as steady state response of the proposed controller. However, power efficiency, computation burden and complexity of algorithm are disadvantage of proposed algorithm.