TSK模糊模型的GA-BP混合学习方法

Learning TSK fuzzy model by GA-BP method

针对TSK模糊模型的学习是多约束和多目标优化问题,提出一种基于GA-BP的TSK模糊模型学习方法。论述了所涉及的相关问题,包括模型结构的种群编码、进化策略及其适应值评估策略,推导了在进化过程中模糊模型前件和后件参数的BP算法。仿真结果表明:该方法具有先验知识要求少、获取的模型具有较好的精确性和简洁性等特点。

Due to the TSK fuzzy model is multi -constraint and multi -target optimization which is difficult to learn, GA - BP hybrid learning method for the model was proposed. Some problems related to a species coding means for the model structure, evolution and fitness evaluation strategy were discussed. The error back propaga- tion algorithm （BP） for training the antecedent and consequent parameters during the process of evolution was derived. The results show that the learning method has the properties of less previous information about objects, better compaction and accuracy for the model.