Abstract
M.Ing. (Electrical Engineering)
Photovoltaic (PV) energy has become the most growing renewable energy source that
serves as an alternative to fossil energy as it is considered cheap, less polluted, etc.
Photovoltaic system works under both uniform irradiance and partial shaded weather
conditions and exhibits both local peak power and global peak power during partial
shaded conditions. Maximum power point tracking (MPPT) techniques are being used
in PV systems to track the local power peak. However, MPPT systems may fail to track
the global peak power. Online MPPT techniques such as Perturb&Observe (P&O) and
incremental conductance (IC) are considered economical and easy to implement.
However, these online methods underachieve due to some flaws such as oscillating
power near maximum power point (MPP) and poor response owing to the sudden
change in irradiance with P&O and IC. MPPT techniques are algorithms used in
photovoltaic (PV) system to extract maximum power from the PV panel.
Offline techniques such as the curve fitting polynomial (CFP) technique use prediction
and fitting method in order to track the MPP. However, lower order CFP does not
extract maximum power from the panel due to inaccurate fitting of the real P-V curve.
The use of supervised machine learning techniques which include artificial neural
network (ANN) and artificial neuro-fuzzy inference system (ANFIS) have improved
the maximum power point tracking (MPPT) process which increased the PV systems
efficiency.
This first contribution of this dissertation presents a reconfiguration approach that uses
series-connected distributive MPPT (DMPPT). DMPPT is used to track the global peak
in order to extract maximum power from the PV array system under uniform irradiance
and partial shaded weather conditions. The second contribution is the implementation
of improved models of online Perturb&Observe and modified incremental conductance
MPPT techniques that work well under sudden change in irradiance, temperature, and
with minimal oscillating power near MPP. The third contribution is the state of the art
of the sixth-order curve fitting polynomial MPPT technique that can extract maximum
power from a PV panel under varied weather conditions. The fourth contribution is to
introduce the use of SVM classifier for optimization and MPPT purpose in a PV system...