Signal processing algorithm of ship IMU accelerometer based on wavelet theory

  Objectives  Based on an analysis of the signal noise characteristics of an inertial measurement unit (IMU) accelerometer, the noise was studied in the wavelet domain.

  Methods  In order to solve the limitations of the traditional Fourier filtering method for nonlinear noise processing such as Gaussian white noise, this paper uses the two aspects of wavelet threshold estimation and threshold processing function to optimize the existing wavelet theory, and proposes a new wavelet threshold denoising algorithm which can simplify the computational complexity of wavelet analysis while effectively improving the performance of small noise reduction. On this basis, the implementation of the new wavelet threshold algorithm on a field-programmable gate array (FPGA) is studied and the system′s performance is tested.

  Results  The results show that the effective resolution of the accelerometer acquisition circuit processed by the new wavelet algorithm can reach more than 22.4 bits, and the effective resolution is improved by 1.6 bits compared with the original signal.

  Conclusions  The results show that the new wavelet algorithm is better than the traditional Fourier filtering and traditional wavelet filtering method in improving the signal-to-noise ratio or reducing the root mean square error, by virtue of which it meets the real-time requirements of signal processing.