Advanced building blocks of power converters for renewable energy based distributed generators

Abstract

The power converter is a critical component of a distributed generation (DG) system, particularly for a renewable energy based distributed generator. With the rapid development and growing applications of DG systems, power converters have evolved from a traditional “power conversion device” to a “system integrator”. This paper outlines the basic and advanced building blocks of power converters for distributed generation systems, with an emphasis on the new algorithms developed especially for grid-connected single-phase power converters in small wind turbines and photovoltaic systems. A grid-connected power converter should feed power of a high quality into a grid with a low total harmonic distortion (THD), and fast dynamic response to the variations of a renewable energy resource. The newly developed space vector based pulse-width-modulation (PWM) with predictive current control provides a direct cycle-by-cycle regulation of the current waveform feeding into the grid. Effective anti-islanding is an essential function of a power converter, in compliance with the grid interconnection requirements. As active islanding detection methods exert disturbances to the grid and passive islanding detection methods generally have large non-detection zones, new and hybrid islanding detection methods have been developed. A power spectral density method has been developed to continuously monitor and identify the possible features of an islanding formation. Maximum power extraction from renewable energy resources is vitally important for distributed generators. Many algorithms have been developed for maximum power point tracking (MPPT) of photovoltaic and wind energy systems. A new MPPT method has been developed by integrating the fast extremum-finding method of Golden Section Search (GSS) into a multi-stage MPPT algorithm. The proposed MPPT algorithm provides fast response (in the order of 40 msec) and high tracking accuracy (99.9%) as compared to the existing algorithms.