中文摘要：

考虑到模糊模型建模精度对于模糊预测控制的重要性，本文提出了一种基于引力搜索超平面聚类的模糊模型辨识方法，并将其应用到水力发电机组模糊预测控制。建立了更具物理意义的超平面聚类模型，结合引力搜索，提出了基于引力搜索的超平面聚类算法（GSHPC），以此辨识模糊模型前提参数，采用正交最小二次法辨识结论参数。在此基础上，进一步提出了水力发电机组的模糊预测控制策略。将本文所提控制策略与传统PID控制以及粒子群优化PID控制进行了对比试验，结果表明基于GSHPC的辨识方法具有较高的辨识精度，本文所提水电机组预测控制方法在提高机绢控制品质方面效果明显。

英文摘要：

This paper presents a novel clustering method for fuzzy predictive control of hydraulic turbine, based on hyper-plane prototype and gravitational search to improve fuzzy modeling accuracy that is crucial to fuzzy predictive control. In this method we developed a hyper-plane prototype clustering model, using gravitational search to construct a gravitational search hyper-plane clustering algorithm （GSHPC）. This algorithm was used for identification of the premise parameters of a T-S fuzzy model, and the consequent parameters were identified with an orthogonal least square method. Then, a fuzzy predictive control strategy for hydraulic turbine was developed using the identification results. A comparison of this strategy with PID and PSO-PID shows that it is very effective in improving control quality of hydraulic turbine and the model enhanced by GSHPC achieves high identification accuracy.