Study on de-noising to the signal of torque sensor in electric power steering system using wavelet transform

According to the characteristics that the signal of torque sensor both have electromagnetic noise of engine and impact signal of road surface, by changing thresholding on the different scales, we present an algorithm that wavelet transform coefficients of noise are filtered and impact signal of road surface are reserved, and reconstruct the torque signal in order to enhance signal-to-noise ratio (SNR) and make the torque signal present enough information of road surface. The experimental results derive from the bench test show that the method is more effective.