Research On Design Optimization Of Power Transformer Based On Genetic Algorithm

Power transformer is one of the important equipments in power generation, transmission, transformation and distribution system, its performance and quality directly determine the power system’s reliability and operational efficiency.The electromagnetic computation is a nonlinear solving process, which is multivariate, multi-constraint and strong coupling. The thesis establishes the mathematical model of transformer, basing on the effective material cost, with design optimization on electromagnetic computation by using genetic algorithm to ensure performance parameters of designed transformer meet the requirements under the most economic manufacturing cost. The effective material cost mainly includes the cost of copper and the cost of silicon steel sheet. At the same time, because algorithm does not deal with constraint conditions, weighted coefficient is applied to the objective function to compensate variables that beyond the constraints in order for the evolution of population and the improved reliability of scheme. Then it is used in the electromagnetic computation process of transformer design with model of SZ11-31500/110, the result shows the algorithm has certain feasibility and validity. At the same time, the optimization measure for on load loss and noise is carried out by analyzing the structure of transformer.