Sheetmaking system identification using synthetic measurement produced

Abstract

In a continuous sheetmaking process, a synthetic measurement is constructed having a higher signal-to-noise ratio than that of each original measurement in situations where more than one (redundant) measurements are available. The input-output dynamic characteristics of each measurement are assumed to be identical except for a gain factor. Using the synthetic measurement, system parameters can be identified in less time with fewer input disturbances. The invention is especially advantageous for application to processes that are expensive to disturb and in which the noise amplitude is high compared to the permissible amplitude of the put disturbance. The signal-to-noise enhanced time-series response measurement may be produced by selecting a time series response measurement that exhibits a greatest magnitude. Alternatively, it may be produced by averaging in the cross-direction a plurality of time series response measurements. Preferably, the enhanced signal-to-noise ratio time-series response measurement is produced, in appropriate circumstances, by taking the root mean square of a plurality of time-series response measurements in the cross-direction. The root-mean-square or average calculation can be conducted at each sample interval such that only the synthetic measurement series needs to be collected in storage but not the matrix of all original measurements.