Tutorial on Niching Genetic Algorithm for Electric Motor Design

摘要

Recently, several non-deterministic algorithms, such as Evolutionary Strategy, Simulated Annealing, Immune Algorithm and Genetic Algorithm (GA), etc., have been applied to electric motor design, in order to satisfy optimally various kinds of design demands for high performance. Among them, GA is known as the most popular one for the design. Genetic algorithms (GAs) are stochastic optimization methods based on the mechanics of natural evolution and natural genetics. They have been successfully applied to finding a global optimum of a single objective problem. In the optimization of multimodal functions, however, the standard GA cannot maintain controlled conpetition among the competing schemata corresponding to different peakd. Moreover, it causes the population to convergo to one alternative or another. In addition, in dealing with a multimodal function with peaks of unequal value, the standard GA converges to the best peak; whereas, in addition to wanting to find the best solution, one may be interested in finding the location of other optima. To overcome these limitations a natural remedy can be tried.