ANFIS训练参数设置研究

Study on The Training Parameters Setting of Anfis

针对自适应神经模糊推理系统的特点,分析了其两种建模方式的优缺点,并最终选择了采用减法聚类算法进行建模.进一步地,针对ANFIS的减法聚类算法4个训练参数初始值的设置问题,采用一多峰函数对其进行了研究,分析了预测结果随该4个参数调节而变化的特点.

According to the characteristics of ANFIS, the shortcomings of its two modelling building methods were analyzed, and the subtractive clustering algorithm was selected finally. Furthermore, the setting method on the initial values of the 4 training parameters of subtraction clustering algorithm was researched by using a multi-peaks function, and the influence of the 4 training parameters was analyzed on prediction result.