Multicomponent decomposition by wavelet modulus maxima and synchronous detection

There are various signal decomposition methods, but none of them is satisfactory and all have their own drawbacks. It is worth exploring a new signal decomposition approach with better performance for processing the complex vibration signals. By employing the wavelet modulus maxima and synchronous detection, a novel multicomponent signal decomposition method is proposed in this paper. Firstly, the wavelet modulus maxima of a multicomponent signal are calculated by optimized complex wavelet transform, then the highest instantaneous frequency (IF) is extracted by searching the wavelet modulus maxima with the smallest scales at all the time instants. With the obtained IF, the synchronous detection method is used to calculate the phase offset and the instantaneous amplitude. It follows that the corresponding component with highest IF can be reconstructed. Then the used wavelet modulus maxima in this iteration are removed from the wavelet scalogram and the next IF is sequentially computed. By repeating this process, all components are successively separated from high frequency to low frequency. Compared with ensemble empirical mode decomposition and Hilbert vibration decomposition, it has been proved by three typical multicomponent signals with different noise intensity that the proposed signal decomposition method has higher accuracy, frequency resolution and is more robust to noises. Moreover, the application results further show that the proposed method can be more effectively applied to fault diagnosis of gearboxes, especially when the operating condition is varying or the fault feature is weak.