Implementation of a predictive modeling technique on a DCS

The implementation of a semi-empirical predictive modeling technique is demonstrated on a level controlled process operated by a distributed control system (DCS). This paper demonstrates the ability of this technique to accurately predict the response of the system with multiple changes in the input variable under a variety of sampling frequencies of the output variable. The new technique was implemented via a function sequence table as part of the configuration software of the DCS. In order for the technique to work, it is necessary to distinguish changes in the input variables over a small sampling interval from measurement noise. Our solution to the problem was analysis and quantification of the noise by use of a threshold to distinguish between real changes and signal noise