Abstract: Combining the de-noising algorithm based on adaptive Gauss filter with the R-statistic method, a steady-state detection was carried out to power plant history data. Taking the total air volume data of a 1 000 MW unit as an object of study, simulation comparison was made between the previously proposed method and the median filter method under same working conditions. Results show that the method based on adaptive Gauss filter has better de-noising effectiveness and larger retention capacity on point mutation information, which is of high sensitivity in short time steady-state detection, and of high accuracy in switching boundary identification between steady-state and unsteady-state detection.