Abstract:
This thesis falls within the framework of harmonic depollution of the low voltage electrical networks using
an active power compensators, this pollution is caused by the unlimited use of non-linear loads. The causes, the
effects of these harmonic disturbances and the international standards imposed to protect the consumers against
these disturbances are presented. Also, several traditional and modern solutions against the harmonic pollution are
presented. We have focused our study on the development of the control techniques of shunt active power filter, For
this reason, we have based on the development of methods for determining harmonic reference currents based on a
multivariable filter (MVF), These new methods are able to correctly extract the reference harmonic currents
necessary to control the SAPF, after that, we have implemented the modified (p-q) method on the power active filter
in order to carry out a global active compensation of the harmonics currents, the behavior and quality of filtering of
the APF are evaluated with two kinds of electric loads, with different types of electrical network and with a
photovoltaic power source. Then, we have developed two harmonic selective identification methods based on
multivariable filter, which are implemented in real time, in order to realise a selective active compensation of
specific harmonic currents under low DC link voltage, this work was closed by an experimental validation of an
active power filter (APF) fed by a PMSG-based wind system, for this we implemented three controllers to control
this system, which are the predictive current controller, Type-2 fuzzy logic controller, and an ANFIS controller. The
experimental results obtained show that this system is capable of compensating for harmonics and reactive power,
and also injected active power into the electrical network.
