In response to the difficulty in extracting early fault feature signals of rolling bearings

a joint fault diagnosis method based on Improved Complete Ensemble Empirical Mode Decomposition (ICEEMD) and Independent Component Analysis(ICA) was proposed. This method utilized the kurtosis criterion to reconstruct the Intrinsic Mode Function (IMF) obtained from ICEEMD and combined it with Fast Independent Component Analysis (FastICA) for noise reduction and unmixing

significantly reducing the noise in the measured signals. The maximum energy amplitude was obtained at the fault feature frequency

making it easy to identify fault features. Through experimental research and analysis

it is shown that this method can significantly reduce noise interference and highlight fault frequency components. Compared with the method combining ICEEMD and envelope spectrum

the signal-to-noise ratio is inereased by 29.54%

which can more accurately identify fault features and meet the discrimination requirements for rolling bearing faults

thus providing a new approach for bearing fault feature extraction.