Research And Application Of Improved Genetic Algorithm In The Design Of Transformer Optimization

The rapid development of economy in China will bring new life and hope to transformer manufacturer. However, it is also a new challenge as the rising of material’s price and the shortening of delivery cycle. The optimization methods and the computer aided design can shorten the design time and reduce the cost to enhance the competitiveness of product in the market and increase economic benefits of product.The thesis studies the optimization of transformer design, establishes the optimization model of transformer based on the Total Owning Cost. According to customer needs and business needs, Total Owning Cost has been integrated as a comprehensive evaluation of the transformer, so purchasers can select the technical feasibility and economic optimal transformer. Objective function adopts Total Owning Cost ideas, because the transformer optimization calculation belongs to strongly constrained optimization problem, applying the thought of penalty function to the objective function for processing, and realizes it can be combined with the thinking of weighted coefficient. Such a way would effective constraint not only the constraints but also in favor of the continuity of genetic evolution and the promotion of population as well.The problem of transformer’s optimization design is a problem of discreteness, multivariable, nonlinear and multi-objective mixed programming. As the faults of prematurity and slow speed of convergence in simple genetic algorithm, it is frequently get the local optimal solution, rather than the global optimal solution. In addition, the simple genetic algorithm can also not meet the dynamic and changing requirements in genetic evolution of certain parameters, in particular, crossover rate and mutation rate. This thesis applies adaptive genetic algorithm to the optimal design of transformer based on Total Owning Cost least, and it is improved in crossover rate, mutation rate and fitness function and several other aspects as to conquer the blemish of genetic algorithm. The results show that the algorithm has good performance, accelerates the convergence speed, improves quality of solution, and helps the optimal design of transformer in processes. It has made meaningful exploration for genetic algorithm in transformer optimization design aspects and further popularization and application.