Inverter design and analysis using multiple reference frame theory.

摘要

Multiple reference frame theory allows for periodically time varying signals to be represented as a set of dc signals. In other words, every periodic signal can be expanded into a Fourier series representation. By modeling an inverter connected to a boost maximum power point tracker (MPPT) in this manner, frequency transfer properties can be preserved and harmonics throughout the system can be predicted. A state space model taking into account the dc and fundamental grid frequency is presented and used to optimize the controller gains of the system. Using information from the dq-axis values of the measured grid current and voltage, the double frequency dc-link voltage component is predicted. The double frequency component is removed from the controller input using feedforward. As a result, there is a reduction in output harmonics in the grid current. The same method is applied to the MPPT, where the double frequency component is predicted and removed from the controller input. This allows for a MPPT with reduced oscillations in the input power waveform. Next, a method is presented to generate a large-signal model of a H-bridge inverter. A set of algorithms are presented, which take a standard set of large-signal (user generated) dynamic equations and performs a Fourier series expansion on the inputs and states of the equations. These algorithms work for an arbitrary finite set of harmonics and preserve the frequency transfer properties between harmonics. The solution to the generated equations is the steady state output of the inverter. Lastly, a set of algorithms are presented which take a user generated netlist in and automatically outputs a truncated harmonic transfer function (THTF).