Abstract: An improved satisfactory clustering algorithm is presented to solve the commonly existed problems in fuzzy clustering algorithm，such as giving clustering number in advance and slow convergence.Using improved satisfactory clustering algorithm，a nonlinear system can be quickly divided into several fuzzy parts，and the sub-models are trained by support vector machine，which are combined by the degrees of membership of input variables.Finally，the industrial simulation shows the efficiency，the accuracy and the rapidity of the method proposed in this paper.

Key words: fuzzy clustering, satisfactory clustering, support vector machine, multi-model

摘要： 针对模糊聚类中普遍存在的聚类个数需要事先给定和收敛速度慢等问题，在原有聚类方法的基础上提出一种改进满意聚类算法。用该算法快速确定系统的模糊划分数目，进而用支持向量机算法建立每个聚类的子模型，将输入变量对各类别的隶属度作为权值，将多个子模型用加权方式组合。工业仿真实例验证了基于该方法的多模型建模方法的有效性、准确性和快速性。

关键词: 模糊聚类, 满意聚类, 支持向量机, 多模型