A robust data reconciliation method for fast metal balance in copper industry

Abstract Data reconciliation along with gross error detection is the key technology for providing accurate and reliable data relating to metal balance in copper industry. However, it can be computationally expensive, especially when the number of variables becomes large, i.e., more than 200, and the constraints are notably complex as in bilinear form. In order to address this problem, a robust estimator-based data reconciliation model for solving the metal balance problem is developed in this study, in which the inconsequent deviation between the measured and reconciled value is fully taken into account, and the gross errors are detected according to the reconciliation results. Specifically, considering the computational efficiency and the early convergence of the evolutionary algorithm, a novel joint optimization strategy is designed to substitute the high-dimensional variables by low-dimensional Lagrange multipliers and restrict the population density in a reasonable range during optimization process to obtain more accurate reconciled results. The practical data collected from a copper plant in China are used to validate the proposed approach. The results demonstrate a significant improvement in performance and computational efficiency with respect to both large-scale data reconciliation and gross error detection thanks to the proposed robust model and joint optimization strategy. Besides, a software system based on the proposed method has been developed and applied in field studies, providing a systematic guidance for practical metal balance.