Using Empirical Mode Decomposition for Underwater Acoustic Signals Recognition

Due to the non-linear and time-varying properties, to process and analyze underwater acoustic signals becomes very difficult and complicated. Fourier transform is only suitable for analyzing stationary signals, but not appropriate for the detection of short time and transient signals embedded in the underwater acoustic signals. Based on the property of multi-scale and multi-translation, wavelet transform can be used for analyzing transient signals. Although the scale and translation parameters in the wavelet basis functions can be adjusted for different signals, the basis functions are fixed, same as the basis functions of Fourier transform. Hence, the requirements of analyzing time-vary underwater acoustic signals are not satisfied. The algorithm of empirical mode decomposition provides a useful analysis scheme for non-linear and non-stationary nature signals. In this work, empirical mode decomposition method is design to extract the features from underwater acoustic signals for recognition. In the experiments, we utilize the data set of different ship classes recorded by the hydrophone to test the proposed scheme. Experimental results demonstrate the robustness of the proposed method.