Fault Diagnosis of Gear Pump Based on Cyclic Impulse Characteristic Decomposition

The information of gear fault features is contained in vibration signals with the form of periodic impulse. Traditional signal decomposition methods mostly are suitable for AM-FM signals. However, when dealing with gear fault vibration signals, the gear impulse feature information is difficult to be accurately extracted from the noise by these methods. Therefore, the cyclic impulse characteristic decomposition (CICD) method is proposed in this article. Firstly, based on the advantage of the periodicity performance of discrete Fourier transform (DFT) matrix, the impulse cycle period is estimated by constructing the cyclic impulse dictionary matrix. Second, as Goertzel filtering method can parallely extract the amplitude information contained in scattered frequency point within different frequency band, and the bilateral spectral amplitudes obtained by this method are reconstructed to get the cyclic pulse characteristic components (CPCCs). The analyzing results of simulation and experiment signals demonstrate that CICD can accurately separate the fault impulse information of gear, as well as has better antinoise ability compared with other methods.