The invention provides a data processing method and device for excitation disturbance signals. The method and device are applied to parameter identification of a synchronous generator, a maximum value sequence and a minimum value sequence in stator current data are obtained through adoption of default extreme point calculation, an envelope curve of the stator current data is obtained on the basis of the maximum value sequence and the minimum value sequence, large signals of gibberish are filtered out, so that the signal waveform of the stator current during excitation disturbance is more obvious. A differential mode is adopted to process exciting current data and power angle signal data so as to obtain a field current changing curve and a power angle signal changing curve, the large signals of the gibberish are filtered out, signal features generated during disturbance are reserved, and reference is provided for follow-up accurate identification of parameters of a generator during on-line disturbance.