With the increasing application of disturbance waveform data in equipment condition monitoring, a disturbance detection algorithm considering the diversity and insignificance of disturbance characteristics is of practical significance in improving the accuracy of equipment initial fault diagnosis. Therefore, based on the characteristics of noise rather than disturbance, a transient disturbance detection method is proposed based on the difference in the probability distribution of residual waveform data. First, the acquisition method of residual waveform data and its probability density is given. Through the regularity analysis of the probability distribution of noise in measured waveform data, the difference of probability density between residual waveform data with or without disturbance is selected as the basis of disturbance analysis. Then the transient disturbance detection algorithm and threshold determination method based on Wasserstein distance between probability density of residual waveform data of each period are established. Simulation and field data verification results show that the proposed disturbance waveform detection method has high detection accuracy for various non-significant disturbances in voltage and current waveform sampling data. This work is supported by the Science and Technology Project of the Headquarters of State Grid Corporation of China (No. 52130421000S).