| With the rapid development of modern industry, mechanical equipment structure isgrowing complicated and automatical, and there are more and more attention about thestate detection and fault diagnosis of the equipment. This paper is aiming at the nonlinearand non-stationary complex features of the mechanical fault vibration signa, proposed afault quantitative analysis method based on fractal theory, finally according to thecharacteristics of complex mechanical failure mode is difficult to identify, given a nuclearfuzzy c-means clustering based recognition method, then apply the above theory toresearch in the swash-plate axial hydraulic pump and rolling bearing fault detection, andthe result is given.First this paper introduces the research background, describes the traditional methodsof vibration signal analysis,including the Fourier transform, Order analysis methods,Neural Network, Fuzzy and Fractal method. This paper describes the advantages of fractaltheory and its application in mechanical fault diagnosis;After highlighting the box method and Multifractal Detrended Fluctuation Analysis,multifractal model is set up, using the simulation signal test and research for the twomethods, summary of main parameters of the multifractal spectrum of discrete data——anomalous scaling exponent α and singular spectrum function f(α) character meaning;This paper, by using the two methods to analysis different data which collected underthe different conditions of hydraulic pump and bearing data, analysis results show that, forhydraulic pump data to Multifractal Detrended Fluctuation Analysis method caneffectively identify the fault signal under the lower sampling frequency, but box methodof multifractal, there is no limit to the sampling condition. Both of them can effectivelyidentify the rolling bearing fault data.Finally, based on two fault feature extraction methods, the use of nuclear fuzzyC-means clustering method to effectively identify the extent of the fault and fault deed. |
PDF Full Text Request