Reconciliation of data with non-random errors

Document Type : Original Article

Authors

1 Department of Mining Eng., University of Kashan

2 Department of Mining Eng., Amir Kabir University of Technology

3 Faculty of Engineering, Tarbiat Modares University

10.22077/jgm.2024.7416.1026

Abstract

Errors of measured data could impact the offline optimizations or online control systems, leading to potentially uneconomical or unsafe process conditions. To address this issue, data reconciliation methods are introduced to enhance the data as much as possible. In this regard, existence of non-random errors is challenging. This article debates the use of conventional sum of squares objective function in the case of presence of non-random errors and shows how a robust estimator such as the maximum likelihood ameliorate the reconciliation. The robustness of the new objective function was assessed using simulated data. Results indicate that the sum of errors between real simulated data of flowrates and their estimation counterparts decreases from 124% to 27% in the case of a gross error in one stream, when robust objective function is manipulated. Even if no non-random error exists, it is shown that robust estimator could result in better reconciliation of data, if optimum parameters are chosen for the robust objective function.

Keywords

Journal of Geomine
Volume 1, Issue 4
December 2023
Pages 171-178

Files

Share

How to cite

Statistics