Date of Award

Spring 1-1-2013

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Electrical, Computer & Energy Engineering

First Advisor

Robert Erickson

Second Advisor

Dragan Maksimovic

Third Advisor

John Hauser

Fourth Advisor

David Meyer

Fifth Advisor

Tom Manteuffel

Abstract

A parallel connected photovoltaic (PV) architecture connects individual PV panels in parallel to a high voltage inverter input bus through intervening dedicated dc-dc converters. These per-panel parallel PV converters provide the large voltage boost normally obtained through the series connection of panels to the inverter input. The goal of this thesis is the development of control algorithms which maximize the robustness and power harvesting of a parallel PV converter, subject to maximum power and current constraints. Meeting this goal requires the invention of several new algorithms and analytical results.

A new state plane technique including generalized diode reverse recovery is developed to produce an accurate model of the parallel converter dc transformer circuit, greatly improving on prior modeling techniques. Existing perturb and observe (PO) maximum power point tracking algorithms are shown to suffer from small local maxima, called traps, in the measured PV power curve. A generalized PO (GPO) algorithm is invented to greatly improve tracking robustness in the presence of traps. A new nonlinear controller is developed which integrates GPO and current limit functions, providing for rapid and stable transition between the two associated modes.

A dead zone avoidance and mitigation (DZAM) processor is invented to address an operational dead zone which exists near the buck-boost mode boundary in a noninverting buck-boost converter. DZAM improves on prior dead zone mitigation techniques for this converter through autonomous operation and the avoidance of any decrease in converter efficiency or operational bandwidth. A significant improvement in the buck-boost efficiency curve is achieved by generalizing the DZAM processor nonlinearity in a way which enables optimal adaptation of an effective converter switching frequency.

Share

COinS