Detection and determination of harmonics by using evolutionary algorithms

Autor: Vlahović-Babić, Josip
Přispěvatelé: Barukčić, Marinko
Jazyk: chorvatština
Rok vydání: 2015
Předmět:
Popis: Kako bi se što bolje uklonile ili ublažile posljedice viših harmonika u sinosuidalnom valnom obliku napona, na električnu opremu i uređaje potrebna je pravilna detekcija i određivanje viših harmonika. Jedan način, na kojem se ovaj rad zasniva, je korištenje genetskog algoritma. Kroz simulacije u programskom paketu Matlab, dobivaju se zadovoljavajući rezultati. Simulirani su utjecaji promjenjive ili konstantne istosmjerne komponente, opisiva ili skokovita promjena amplituda harmonika, utjecaj međuharmonika te šuma u signalu. Rastavljanje ukupnog problema na dva manja problema te pokretanje dva genetska algoritma, po jedan za svaki problem, se pokazalo puno uspješnije nego rješavanje cijelog problema jednim genetskim algoritmom. Najbolji rezultati za ovakav problem pokazali su parametri: turnirska selekcija, adaptivno izvediva mutacija te raspršeno križanje. Harmonijska analiza genetskim algoritmom se isplati koristiti kada se radi o kompliciranim signalima jer je genetski algoritam moguće prilagoditi raznim problemima. In order to eliminate or mitigate the consequences of higher harmonics on the electrical equipment and appliances it is important to conduct proper detection and determination of the higher harmonics in the sinusoidal wave form. This work is based on one particular way of doing this, using genetic algorithm. Satisfactory results where obtained through simulation in Matlab software package. The effects of a variable or a constant DC component, mathematically describable or step changes of the amplitude of harmonics, inter-harmonics and the impact of noise in the signal where simulated. Dismantling oft he overall problem into two smaller problems, and running two genetic algorithms, one for each problem, turned out a lot more successful than solving the whole problem with one genetic algorithm. The best results for this kind of problems where obtained using the following parameters: tournament selection, adaptive feasible mutation and scattered crossover. It is worthwhile using genetic algorithm when dealing with complicated signals because the genetic algorithm can be adopted to suit various problems.
Databáze: OpenAIRE