In regard to stochastic and suddenly varying data

a new kind of data processor algorithm based on adaptive prediction is presented.The algorithm describes the changing pattern of measured variables by means of time series models with slowly time-varying parameters

forms the criterion by use of the adaptive and its 95% belief limit

and processes the measured data so as to reject or restrain outliers mixed in them.The simulation example indicates that the algorithm surmounts difficulties which most current algorithms encounter when processing stochastic and suddenly varying data and that it is of obvious effects.