The measurement accuracy and reliability of the procedure parameters of thermal power plants are the decisive factors that restrict the application of SIS. This paper describes two data-repair processing methods including multipoint and single point conditions. Based on the probability theory and the weighted method， the multipoint-data processing method is modified. According to the consecutive mutative rule of a single measuring point， an extrapolation scalping method is provided， in which the abnormal values are replaced with the fitted ones to improve the reliability of procedure parameters. The two simple data processing methods of high efficiency and reliability are successfully applied in SIS of a power plant.