Designing fuzzy controllers with variable universes of discourse using input–output data

• An improved recursive least square algorithm is presented to design VUD fuzzy controllers using subconscious knowledge.
• A cut-off condition is used to end the identification process.
• It is feasible to use the ergodic property of input–output data to get the cut-off condition.

Using field data from sensors to construct a fuzzy controller based on variable universe of discourse (VUD) is an open question that is often encountered in real-world applications. The basic structure of a VUD fuzzy controller is given in this paper. To use subconscious knowledge, the input/output data from sensors, a novel method is proposed for designing a VUD fuzzy controller. Firstly, the central points of the input fuzzy sets are distributed equidistantly and those of the output fuzzy sets are regarded as undetermined parameters. Then, an improved recursive least square algorithm is presented to identify these unknown parameters, and a cut-off condition is applied to end the recursive computation. Finally, simulation examples show that the VUD fuzzy controller constructed by the identification algorithm can satisfactorily approximate original input/output data and that it can achieve a better effect on a rotary single inverted pendulum than a general fuzzy controller.