This thesis focuses on the study of methods for harmonic compensation in electrical power systems and discusses various aspects related to the presence of harmonics on them, such as the presentation of concepts and definitions in nonsinusoidal systems and strategies of power compensation. Exemplified by measurements and simulations the following topics are emphasized in this study: the influence of supply waveforms on nonlinear loads; the harmonic interaction between the supply voltage and the load currents, due to the system series impedance; and the mutual influence between nonlinear shunt loads, as a possible way of harmonic reduction. To simulate and predict the impact caused by nonlinear loads in a system, as well as the implementation of actions to mitigate these impacts in order to improve power quality, it is necessary to know the response of these loads. As a product of the present work, the development of techniques for modeling nonlinear loads under different supply conditions is emphasized, especially those using computacional intelligence, such as the neuro-fuzzy system and the artificial neural network; as well as the use of Volterra series for load behavior prediction.
|
