Repetitive impacts recovering using variational mode extraction with constructed reference enhanced by improved blind deconvolution

Abstract

The impulse and modulation characteristic of rolling bearing’ faulty signal is often very weak when early fault arises in rolling bearing or gears, and the main reasons are due to the signal attenuation caused by too long signal acquisition path and the interference of other multi-source vibration. In order to extract the weak feature accurately, a method named as variational mode extraction (VME) based on constructed reference enhanced by improved minimum entropy de-convolution (IMED) is proposed, which combines both the advantages of IMED in solving the influence of the long signal acquisition path and VME based on constructed reference in extracting the impulse and modulation characteristic of vibration signal. Firstly, IMED is used as signal preprocessing method to analyze the vibration signal of rotating machinery to eliminate the influence of long signal acquisition path and enhance the repetitive impulse characteristics. Then, reference signal is constructed according to the prior knowledge of the rotating machinery and input it with the output signal of IMED into the VME model together, and the output result of VME not only could further enhance the impulse characteristic of vibration signal, but also obtain the modulation characteristic simultaneously. Finally, envelope spectral or enhanced envelope spectral is performed on the output signal of VME and satisfactory fault features are extracted. In order to solve the shortcomings of traditional MED, an IMED based on D-norm is proposed which has higher computational efficiency and could extract multi-harmonic impulse features. In addition, VME based on constructed reference is proposed to improve the accuracy of VME in extracting the target signal. Feasibility and superiority of the proposed method are verified by one experimental case and one engineering case.