Robust kernel adaptive filtering for nonlinear time series prediction

Long Shi, Jinghua Tan, Jun Wang, Qing Li, Lu Lu, Badong Chen

Abstract: Highlights•The kernel adaptive filter with generalized tanh function (KAF-GT) is proposed, which shows favourable performance in combating impulsive noise.•The convergence analysis of the proposed KAF-GT algorithm is conducted using energy conservation relation.•By using the pre-tuned dictionary strategy to model the unknown nonlinear system, we perform the EMSE analysis and arrive at a closed-form solution for predicting the steady-state behavior.•To further improve the performance of KAF-GT, an optimization scheme for scale factor is designed.