Designing a fuzzy model by adaptive macroevolution genetic algorithms

Abstract

In this paper the adaptive macroevolution genetic algorithms are proposed to identify three different types of fuzzy models. Several newly established techniques, such as adaptive choice function and macroevolution, are adopted into the simple genetic algorithms to improve the optimization capability. The genetic algorithms used here are controlled to retain the best solution in the population until a better one is found in the next generation. The procedures of applying the proposed genetic algorithms to optimize the fuzzy models are also investigated. We also compare whether the instantaneous or batch model is more suitable to adjust the fuzzy models. The refined fuzzy models are then used to remedy the prediction output from a grey system. The superiority of the adaptive macroevolution genetic algorithms to the simple ones is discussed and an example is given to verify our viewpoints. Simulation results from difference equation and prediction filter are presented to illustrate the effectiveness of the proposed genetic-based fuzzy grey model.