Plant-wide Mass Balance using Extended Support Vector Regressionbased Data Reconciliation and Gross Error Detection

摘要

In any modern petrochemical plant, the plant-widemass data rendering the real conditions of manufacturing is thekey to the operation managements such as production planning,production scheduling and performance analysis. Because of thecharacteristic of data reconciliation and gross error detection, itis quite suitable to address plant-wide mass balance problemusing data reconciliation and gross error detection techniques.In this paper, an extended support vector regression approachfor data reconciliation and gross error detection is proposed toachieve plant-wide mass balance, which can simultaneouslydetect and estimate measurement errors' and missing massmovement information. The simulation results demonstrate thatthe proposed approach is effective and accurate.