The structural identification as a proposed of classification of variables for the reconciliation of industrial data

Abstract

Data reconciliation is strongly affected by formulation problem, statistical results interpretation and optimization performance. This must be valued by a carefully variable classification. Due to of the complexity of integrated process and the large volume of available data in highly automated plants; classification algorithms are increasing used nowadays. They are applied to the revamps, design and monitoring systems and to reduce the dimension of the data reconciliation problem. A new algorithm is described which can help the engineer find efficient strategies for the classification problem allied with the mathematical formulation. Some structural properties are discussed and illustrated. The new structural identification algorithm is described. There is a large economical incentive for the variable robust classification, because a defective procedure will request an additional instrumentation.