On-Line Reconciliation Of Mineral Processing Data

Computer systems, installed in mineral processing, are designed to accumulate massive amounts of real-time data. Frequently these data are only partially processed because of the lack of adequate methods and the limited accuracy of mineral processing sensors. Methods are proposed to estimate missing data and reconciliate sensor data on a real-time basis. They are based on dynamic extension of steady-state techniques such as the node imbalance minimization or the least-squares mass balance adjustment techniques. The methods are illustrated for simulated and actual flotation data. Their data upgrading ability could be used to improve process monitoring, control and optimization.