Wavelet Broad Learning Filter: A Novel Adaptive Filter for estimating the Physiological Tremor in Teleoperation

Abstract

In this paper, a wavelet broad learning filter is proposed to estimate the tremor. At first, the structure of original broad learning system (BLS) is redesigned. To extract the features of tremor, the novel WBLAF maps each dimensional data as the feature nodes respectively by the wavelet function. Secondly, a novel self-paced wavelet auto-encoder (SPWAE) is proposed to train the weights of feature mapping. In addition, the ridge regression learning algorithm and the incremental learning of the proposed filter are applied to learning online. Finally, semiphysical simulation experiment is accomplished. As shown in the results, the new proposed WBLAF can effectively estimate and filter out the physiological tremor in tele-operation.