Identification of bad data is accomplished by removing
measurements which differ from the mean value by greater than preset tolerance values respectively, where the values denote the sample standard deviation values of , and as measured in a moving window of width .