Data Reconciliation with Simultaneous Bias and Leak Estimation Based on Generalized T Distribution and Akaike Information Criterion

Liyong Xiao,Yu Miao,Hongye Su

2011-01-01

Abstract:A data reconciliation with simultaneous bias and leak estimation approach is proposed in this paper, which is based on combining merits of the generalized T distribution method and the extended Akaike information criterion (AIC) method proposed in this paper. This approach makes use of GT distribution function to eliminate the effects of measurement biases and applies extended AIC approach to address process leaks to achieve accurate data reconciliation and estimate measurement biases and process leaks on even nonlinear steady systems. This combination will retain the advantage of robust estimator to adaptively fit to measurement errors distribution and will also consider process model uncertainty such as process leaks. The Simulation results from a heat-exchange network, a nonlinear steady system, demonstrate the accuracy and effectiveness of the proposed approach.