FAST ONLINE DATA EVALUATION OF FLOW-INJECTION ANALYSIS SIGNALS BASED ON PARAMETER-ESTIMATION BY AN EXTENDED KALMAN FILTER

The present paper is concerned with the fast evaluation of the flow in jection analysis (FIA) signals and the automatic correction of the ana lytical values interfered by systematic and stochastic disturbances. W ith the application of the extended Kalman filter, the highest amount of information for the data evaluation of analytical signals can be ex tracted fi om FIA peaks. The concentration of the analyte and the offs et of the baseline are estimated as time-variable parameters by filter ing. The results of the application to simulated and real FIA data sho w that the parameters corresponded to the evaluation of the analytical data, but is already available before the maximum of the FIA peak, i. e, the evaluation of the data with the Kalman filter can be done durin g the running peak. Good estimates of the measured values are already obtained short after start of the peak. For this reason, the measuring dead-time of the FIA system can be reduced by the use of this method. The evaluation of FIA signals disturbed often by fluctuation of the b aseline and other noise can be corrected by estimation of the offset o f FIA peaks and smoothing of real FIA signals. Therefore, the applicat ion of the extended Kalman filter can also improve conventional evalua tion methods. (C) 1998 Elsevier Science B.V. All rights reserved.