Traditional singular data identification and correction methods process data roughly and cannot accurately deal with the shortcomings of singular data, this paper proposes a singular data identification and correction method based on wavelet analysis, which uses the localization properties of wavelet analysis in terms of time domain and frequency domain with the "micro" features of signals. First of all, wavelet analysis is conducted to extract the high frequency component signal as well as characterization of random noise, combined with probability statistical method to analyze the high frequency component signals, determine the occurrence time of singular data and finally eliminate singular data. The linear interpolation method is used to supplement the correction. A large number of examples show that the method is correct and effective.
