Genetic algorithm for design optimization of 3-phase semi-conductor rectifier power transformers

This paper presents the optimal design of a 3-phase rectifier power transformer supplying a dc load by Genetic Algorithm (GA). A closed form expression for actual phase current waveform of the transformer is derived and a comparison is made with the approximate waveform normally considered for fixing the transformer rating. Powell's direct search method as optimization technique is applied for the estimation of transformer design parameters and compared with the results obtained using GA. The optimal results demonstrated by an example show the potential for implementation of GA as an efficient search technique for design optimization of transformers. A discussion on the limitations and variation of GA parameters while minimizing the single and multi-objective functions satisfying the performance constraints on the proposed optimal design concludes the paper.