Accelerate Convergence of Polarized Random Fourier Feature-Based Kernel Adaptive Filtering With Variable Forgetting Factor and Step Size

Abstract

The random Fourier feature as an efficient kernel approximation method can effectively suppress the network growth of the traditional kernel-based adaptive filtering algorithm. Polarized random Fourier feature kernel least-mean-square(PRFFKLMS) remarkably improved the accuracy performance of random Fourier feature-based kernel least-mean-square algorithm and become the most representative random Fourier feature-based least-mean-square algorithm. In this paper, we studied the method that can improve the convergence speed of random Fourier feature-based least-mean-square algorithm. Based on the variable forgetting factor and variable step size strategy, three algorithm are proposed. The computational complexity of proposed algorithms are also given. The simulation results show that compared with PRFFKLMS algorithm, the convergence speed of the proposed algorithm is significantly improved. INDEX TERMS Random Fourier features; forgetting factor strategy; variable step size; kernel least-meansquare algorithm.

An error covariance correction-adaptive extended Kalman filter based on piecewise forgetting factor recursive least squares method for the state-of-charge estimation of lithium-ion batteries

“…When the forgetting factor is small, the historical data are easily forgotten to a large extent, and the tracking output changes are better, but the stability is poor. Therefore, the forgetting factor is generally between 0.95 and 0.995 according to previous research [54]. To sum up, a single forgetting factor cannot precisely implement the model identification process in a system with time-varying parameters.…”

Variable Forgetting Factor-Based Adaptive Kalman Filter With Disturbance Estimation Considering Observation Noise Reduction

“…Therefore, this paper uses a variable forgetting factor (VFF) with an adaptive law, which considers the balance between disturbance compensation performance and noise sensitivity, to design the AKF. In developing recursive least squares (RLS) methods for unknown parameter estimation, the VFF [22]- [24] has been used [25]- [28]. The VFF-based AKFD can adequately handle observation noise and provide an accurate simultaneous estimation of state and disturbance variables.…”