Rotating machine fault diagnosis based on intrinsic characteristic-scale decomposition

• A new self-adaptive time–frequency analysis method ICD is proposed.
• ICD method integrates the advantages of LMD and ITD.
• ICD can avoid some limitations existing in LMD.
• Experimental analysis proved the feasibility and effectivity of the proposed method.

A new method called intrinsic characteristic-scale decomposition (ICD) is proposed in this paper, which is particularly suitable for processing the nonlinear and non-stationary time series. When fault occurs in gearbox and rolling bearing, the measured vibration signals would exactly present non-stationary characteristics. ICD, a new self-adaptive time-frequency analysis method, can decompose the non-stationary signal into a series of product components (PCs). Therefore, it is possible to diagnose gearbox and rolling bearing fault. In this paper, the ICD method is introduced and the decomposition performance is compared with LMD method. The results demonstrate that ICD has the advantages at least in running time, alleviating the mode mixing problem and restraining the end effect. The ICD method is applied to the practical gear and rolling bearing fault diagnosis. The results demonstrate that the proposed method is effective in the fault signature analysis of the rotating machinery.

The local mean and envelope estimation of ICD is more approximate to the standard m(t) of x(t),which means the ICD method can improve the envelope approximation accuracy of local mean and get more accurate results than LMD method.A comparison of the normalized local mean function of ICD and LMD with the standard local mean of x(t).Figure optionsDownload as PowerPoint slide