基于超平面原型聚类的水轮机调速系统模糊模型辨识  

Fuzzy Model Identification of Hydro-turbine Governing System with Clustering Based on Hyperplane Prototype

作　　者：李超顺[1] 周建中[1] 向秀桥[1] 刘力[1] 贺徽[1] 张勇传[1] 

机构地区：[1]华中科技大学水电与数字化工程学院,武汉430074

出　　处：《动力工程》2009年第4期363-368,388,共7页Power Engineering

基　　金：科技部水利部公益性行业科研专项资助项目(200701008);国家自然科学基金雅砻江联合研究基金重点资助项目(50539140);高等学校博士学科点专项科研基金资助项目(20050487062)

摘　　要：针对水轮机调速系统的辨识难题,提出了1种基于超平面原型聚类的T-S模糊模型辨识方法.基于局部模糊模型线性度的重要性,推导出1种基于超平面的模糊聚类算法.该算法以优化局部模型线性度为目标,进行模糊模型前提结构辨识,能使局部模型具有良好的线性度;它应用变尺度混沌优化方法搜索最优聚类结果,避免陷入局部极小;应用最小二乘法实现模糊模型结论参数辨识.以某水电厂水轮机调速系统为对象,采用该方法建立了T-S模糊模型,并对其进行了辨识和对比试验.结果表明:建立的T-S辨识模型具有较高的辨识精度及较强的泛化能力,提出的模型辨识方法有效可行.A new Takagi-Sugeno （T-S） fuzzy model identification method based on hyperplane prototype clustering was proposed to solve the identification difficult of hydro-turbine governing system. Considering the importance of local models linearity, a fuzzy clustering algorithm with hyperplane prototype was deduced. Aimed at optimizing linearity of local models, this algorithm provides good linearity in structure identification of premise parts of T-S fuzzy model. This algorithm adopts mutative scale chaos optimization strategy to search （LSM） is used to built for hydro-tur the best clustering results, which can avoiding local optimum. Least Square Method identify fuzzy model consequent parameters. By this method, a T-S fuzzy model was bine governing system of hydropower plant. Results of identification and comparison experiments for this model show that, the T-S fuzzy model has high identification accuracy and strong generalization ability,and the identification method is available and feasible.

关 键 词：水轮机 调速系统 系统辨识 T-S模糊模型 超平面 模糊聚类 混沌优化 

分 类 号：TM312[电气工程—电机] TP18[自动化与计算机技术—控制理论与控制工程]