Abstract: A novel engine fault diagnosis method based on volterra mode of variational mode decomposition (VMD) and morphology fractal dimension is proposed to solve the problem of the nonlinearity of the measured engine fault signal and that morphology fractal dimension can not estimate this signal effectively. Firstly, the engine fault signal is decomposed by VMD method, and the noise and false interference components are separated from the signal by the component elimination algorithm of the false intrinsic mode function (IMF) based on mutual information entropy and energy entropy increment. The sensitive IMF components with fault information are reconstructed. Then the Volterra adaptive prediction model is established by reconstructing the phase space of reconstructed signal to obtain the model parameters. Finally, the morphology fractal dimension of the model parameter matrix is calculated and these characteristic parameters are used to identify engine working states and fault types. Through analysis of the measured engine sound vibration signals at different states, the experimental results show that the proposed method can effectively extract the state information characteristics of the engine and realize fault diagnosis for engine abnormal sound signals.

Key words: engine, fault diagnosis, variational mode decomposition, Volterra prediction model, mathematical morphology, fractal dimension