Successive difference method for steady state detection in univariate system

The use of online data for steady-state detection is required to solve problems like statistical data reconciliation, real time optimization and controller performance monitoring. In this paper, a new method for univariate system is proposed, which makes use of successive differences of time series (single variate) data. The method is simple because the parameters on which it is based are easy to tune as they are rather intuitive. Also this method needs less computation time as it does not involve any model fitting exercise as the case with other methods like polynomial interpolation technique. In order to assess the performance of the above method, a comparison analysis based on its performance in accurately detecting the steady state part that is present in a set of industrial time series data was performed. The performance of this method is compared with the three best existing methods that are available in the current literature. This analysis showed that the proposed method, Successive Difference method is most robust and its performance is better than the existing three methods.