A study of nonlinear dynamic data reconciliation

Data reconciliation is based on spatial redundancy to adjust process data to improve the quality of measurement corruption due to measurement noise. Based on the analysis of characteristic of moving window, we proposed in this paper to use temporal redundancy in nonlinear dynamic data reconciliation to improve accuracy of reconciled value. The reconciled value by using the proposed approach is closer to true data compared with the reconciled value by using the conventional method. The mathematical formulation of outlier detection strategy is given in this paper also. Simulation results on a continuous stirred tank reactor (CSTR) process verified its effectiveness.