Statistical Learning Modeling for Tremor Signal Based on Empirical Mode Decomposition Method

Modeling and analyzing the human tremor signal is necessary to avoid its negative effect for the fine operation. However, there are some defects in the traditional method for tremor signal analysis, which cannot resolve the localization contradictions in time domain and frequency domain. This paper proposes the statistical learning modeling method for tremor signal, which decomposes the tremor signal based on the empirical mode decomposition method, and constructs a composite two-order linear model for tremor signal based on the recursive least squares method with forgetting factor. Simulation results showed the high accuracy of the tremor model, which will be used to filter out the tremor signal during fine operation and improve the precision and stability of the operation.