Basic Statistical Tests For Gross Error Detection

The technique of data reconciliation crucially depends on the assumption that only random errors are present in the data and systematic errors either in the measurements or the model equations are not present. If this assumption is invalid, reconciliation can lead to large adjustments being made to the measured values, and the resulting estimates can be very inaccurate and even infeasible. Thus it is important to identify such systematic or gross errors before the final reconciled estimates are obtained. In the first chapter, it was pointed out that reconciliation can be performed only if constraints are present. The same statement can be made with regard to the detection of gross errors. Without the availability of constraints as a counter-check of the measurements, gross error detection cannot be carried out. Thus both data reconciliation and gross error detection techniques exploit the same information available from measurements and constraints. These techniques, therefore, go hand in hand in the processing of data.