Data Reconciliation Based on Large Scale Rigorous Model with Equation Oriented Method

张正江,祝铃钰,邵之江,陈曦,钱积新

DOI: https://doi.org/10.3321/j.issn:1003-9015.2008.05.027

2008-01-01

Abstract:Reliable process model and process data are required in the real-time optimization. As a result of random and possible gross errors existing in the measured process data, the real-time optimization is not efficient and accurate. According to the characteristics of process system, data reconciliation based on large scale rigorous model was proposed in this paper. The membership functions of gross error and random error were constructed and used to detect the gross error. When there is plenty of measurement information, the reconciliation is available for the process measurement of flow rate, components, pressure, etc. The proposed method for data reconciliation and gross error detection was used in the ethylene separation process system. If only random errors exist in the process data, the reconciled data satisfy the process model. If both random and gross errors exist in the process data, the gross errors can be detected accurately, and the process model is also satisfied after data reconciliation. The effectiveness of the method proposed was demonstrated by the simulation calculation.