关键词: 循环谱负熵, 解卷积, 滚动轴承, 故障诊断, 广义瑞利熵, 区间窄带技术

Abstract: The deconvolution method is an effective mechanical fault diagnosis method,which can reduce the noise and eliminate the interference of the transmission path of the system. However,most of the complex mechanical equipment are composed of multiple rotating parts,and the impulses excited by the failure of the bearings of the key components are often drowned in the Gaussian noise and multiple periodic impulses,which makes it difficult to accurately extract the characteristics of the faults with the traditional deconvolution method. In order to overcome the problem,a rolling bearing fault diagnosis method based on cyclic spectrum negative entropy deconvolution was proposed. Firstly,the original signal frequency band was divided equally,and the initial filter was determined according to the envelope spectral gradient of each frequency band to guide the direction of the iterative process. Secondly,the initial filter was optimized iteratively by the cyclic spectrum negative entropy-based feature vector method. Finally,the original vibration signal was filtered by the optimized filter and subjected to the envelope analysis, the fault diagnosis of complex mechanical was achieved. Simulation and experimental results show that the proposed method can effectively suppress the interference of natural periodic impulses and ambient noise,and has stronger generalization ability and diagnostic performance than the traditional deconvolution method.

Key words: cyclic spectrum negative entropy, deconvolution, rolling bearing, fault diagnosis, generalized rayleigh entropy, interval narrow-band technique