Hide/Show Apps

High performance current control methods for voltage source converters with saturable inductors

Özkan, Ziy
Pulse Width Modulated (PWM) Voltage-Source Converters (VSCs) are the building blocks of today’s power electronics technology for grid-connected systems. In PWM-VSC systems, utilization of filter inductors with deep saturation characteristics is often advantageous due to improved size, cost, and efficiency. However, with the inclusion of saturable inductors, current control dynamics become nonlinear. The utilization of conventional linear current regulation methods in such nonlinear systems results in significant dynamic performance loss as bandwidth shrinkage and poor steady-state current waveform quality as amplification of low-frequency harmonics. This thesis proposes an Inverse Dynamic Model Based Compensation (IDMBC) method and a Saturation Compensation with Resistive Decoupling (SCRD) method to overcome the performance issues of PWM-VSC systems with conventional linear current regulators. Prior to implementation, the nonlinear VSC system L-R parameters are characterized. Employing the nonlinear system characteristics foreknowledge, the proposed methods linearize the nonlinear plant in the large yielding a fictitious linear plant such that linear controllers perform satisfactorily. These methods can be applied to single-phase (1P) (half-bridge (HB) and full-bridge (FB)) and multiphase HB and FB VSC systems. In 1P (HB/FB) and multiphase HB systems single-line dynamic parameters are used in the linearization. Unconventionally, in multiphase FB systems the line-to-line nonlinear system model is established and utilized in the linearization. The proposed methods are applied to single-phase and three-phase PWM-VSC grid-connected systems. A thorough dynamic response, parameter mismatch, and steady-state performance assessment of the methods is performed in comparison with conventional methods. Advantageous performance attributes are demonstrated via analyses, simulations, and laboratory experiments.