Stress wave signal denoising using ensemble empirical mode decomposition and an instantaneous half period model.

Stress-wave-based techniques have been proven to be an accurate nondestructive test means for determining the quality of wood based materials and they been widely used for this purpose. However, the results are usually inconsistent, partially due to the significant difficulties in processing the nonlinear, non-stationary stress wave signals which are often corrupted by noise. In this paper, an ensemble empirical mode decomposition (EEMD) based approach with the aim of signal denoising was proposed and applied to stress wave signals. The method defined the time interval between two adjacent zero-crossings within the intrinsic mode function (IMF) as the instantaneous half period (IHP) and used it as a criterion to detect and classify the noise oscillations. The waveform between the two adjacent zero-crossings was retained when the IHP was larger than the predefined threshold, whereas the waveforms with smaller IHP were set to zero. Finally the estimated signal was obtained by reconstructing the processed IMFs. The details of threshold choosing rules were also discussed in the paper. Additive Gaussian white noise was embedded into real stress wave signals to test the proposed method. Butterworth low pass filter, EEMD-based low pass filter and EEMD-based thresholding filter were used to compare filtering performance. Mean square error between clean and filtered stress waves was used as filtering performance indexes. The results demonstrated the excellent efficiency of the proposed method.