07. Knowledge and Information Management
The adjustment of data obtained in an experimental procedure so that they meet process constraints, such as mass and energy balances, is known as data reconciliation. In some cases, simultaneously to the reconciliation, it is necessary to estimate the parameters of a process model. The simultaneous evaluation of data reconciliation and parameter estimation involves an optimization problem build on statistical hypotheses. It is possible to use the reconciled quantities and the estimated parameters the optimization of process plants or statistical process control, so it is important to evaluate the statistical significance of the results obtained, which depends on the validity of the assumed hypotheses for the optimization problem. Thus, this work assesses the impacts of the assumptions adopted for the simultaneous evaluation of data reconciliation and parameter estimation, in addition to the statistical interpretation of the results especially the behavior of residuals, through a case study applied to a heat exchange process.
PALAVRAS-CHAVE: data reconciliation, parameter estimation, residual analysis, weighted least squares