In order to solve the problem that the measured axis trace is polluted by noise and can not reflect the real fault type of the turbine generator rotor, the adaptive local iterative filtering is applied to the purification of the axis trace of the generator rotor. As an adaptive mode decomposition method, adaptive local iterative filtering can adaptively decompose a complex original signal into a series of single components reflecting the wave nature of the signal, then, by calculating the correlation coefficient between the autocorrelation function of each decomposition component and the autocorrelation function of the original signal, the real intrinsic mode function is extracted, finally, the extracted intrinsic mode function is used to synthesize the signal to purify the axis trace. The simulation analysis and the purification results of the measured axis trace show that the adaptive local iterative filter can effectively avoid the false components in the decomposition process and is more favorable for extracting the signal features by introducing the Fokker-Planck equation to design the filter, compared with ensemble empirical mode decomposition, this method has better purification performance.