Roll-learning algorithm for fuzzy logic system based on genetic algorithm
Deng Jian · Kongzhi yu juece · 2002
A local optimization strategy is presented as learning method for fuzzy logic system. It generates fuzzy rules from the input output data pairs in a local area and trains the parameters of these rules using differential evolution method. To prevent from influencing the performance of fuzzy logic system in adjacent areas, a technique using roll updating data window is used, which involves not only the data in the current local area but also the data in neighborhood. This algorithm reduces the computation and makes it possible to use genetic algorithms in online learning of fuzzy logic system.