Analysis and Design of an Efficient PV Power Optimizer with Reduced EMI Effects and Less Sensors Counts

Abstract

Power optimizers (POs) based distributed maximum power point tracking (DMPPT) architecture can reduce the shading and mismatches losses in photovoltaic(PV) systems. The POs are inherently switched dynamical systems characterizedby discrete switching events that make the systems toggle between two or moretopological states in local neighborhood of the maximum power point (MPP). Due
to this switching process, MPP tracker results in various harmonics at the multiplesof switching frequency of the power converters. These harmonics are undesirable as they often associated with electromagnetic interferences (EMIs) and may also degrade the performance of the system; in particular, for DMPPT architectures. The MPP trackers must provide the desired electrical functionality, e.g., meet the EMIs regulation-standards and also exhibit the fast-tracking performance under rapidly changing solar irradiation and load uctuations.
The aim of this work is thus being devoted to the development of rapid and precise MPP trackers for various PV applications. However, constrains imposed by cost, number of sensors requirement, size/weight, and tracking performances (in terms of fast transient responses with bounded chaotic ripple speci_cations in order to reduce EMIs) essentially limit the application of conventional control techniques and their analysis methodologies. We propose an analog MPP tracker which is deliberately designed and analyzed by using the concepts of nonlinear dynamics and bifurcation theory. Such concepts not only provide the informations to design a fast-and-e_cient MPP tracker under rapidly changing environmental conditions, but also guarantees the system to operate in chaotic mode. We have developed both 1-D and 2-D discrete-time models (or maps) that will ensure reliable and safe chaotic operation of the modular photovoltaic systems (MPVS). The conditions for robust chaos thus also been derived and show how its robustness can be destroyed by introducing a small change in the switching control logic, e.g., from synchronous to asynchronous mode of operation and vice-visa. In addition, an exemplary concept of current sensorless MPP tracker with two-loop (i.e., fast inner peak current-mode-controller and slow outer voltage controller) feedback control technique has been proposed. The appropriate design of such MPPT tracker has also been performed using the singular perturbation based fast-slow-scale system analysis. We show that use of current sensors for MPP tracker can be
completely avoided and exact current information can be estimated by using a proportional-integral observer (PIO). The advantages of such PIOs in terms of ability to estimate simultaneously the states and the unknown inputs disturbances/model uncertainties have been explored and then compared with classical Luenberger observer or P-observer. Finally, all these are experimentally veri_ed using a built-in laboratory prototype MPVS.