A New Clustering Analysis Used for Generating Fuzzy Control Rules

Due to the nonlinear and time-variant characteristics of a complicate industrial process, modern industrial control engineers are more likely to adopt fuzzy control method. In a fuzzy controller, its core is a group of fuzzy control rules. This paper presents some data mining methods to generate the language variables on which fuzzy control rules´ generation depends. These methods are incremental fuzzy clustering algorithm in a batch mode, data weight associated with scale of samples control, divisional threshold value control. These methods can achieve the targets as follows: (1) update the rule base dynamically: (2) limit the samples´ infinite growth and adjust controlled model dynamically by real-time information; (3) control the object exactly and quickly.