中文摘要：

针对模糊建模在进行结构辨识时需事先设定聚类数的问题,本文在改进模糊分割聚类算法的基础上,对算法中聚类数c给出优选方法,提出了参数自适应模糊聚类算法,并结合递推最小二乘法构建T-S模糊辨识算法。为了验证本文提出的模糊辨识方法的有效性,采用该算法对熟知的Box-Jenkins煤气炉数据和实际的电液位置伺服系统数据进行建模,结果显示该辨识方法具有较高的逼近精度和较好的泛化能力。

英文摘要：

Focused on the request of pre-setting the number of clusters for the structure identification with fuzzy modeling. On the basis of the improved fuzzy clustering algorithm, an optimal selection method for the clustering number c and the parameter adaptive fuzzy clustering algorithm are proposed. Combined with the recursive least squares method, a T-S fuzzy identification algorithm is presented. Finally, for demonstrating the effectiveness of the proposed method, the well-known Box-Jenkins gas furnace data and the actual direct drive electro-hydraulic position servo system are identified, respectively, and the identified results show that this identification method has higher accuracy and better approximation capability.