In this paper, a partially recurrent fuzzy systemis developed to function as an adaptive noise canceller. In order to cancel noise distorting the information signal, the temporal information (dynamics) underlying the noise source and the distorting noise, which is generated by the noise source passing through some unknown channels, should be captured accurately. For this purpose, short-term memory is embedded into the input layer of the fuzzy system for handling local time information and internal feedback is introduced into the consequent part for processing global time information by virtue of a partially recurrent mechanism. A novel adaptive algorithm is proposed to tune the parameters of the premise and consequent part online. Simulation studies show that the proposed fuzzy system can cancel noise cancellation successfully for nonlinear dynamic channels.
展开▼
