Intelligent Optimal Framework for the Industrial Mining Plant-Wide Prediction Control

Abstract

In the underground mining industry, the deep cone system is the core element for the whole production and paste-filling process. Due to the coupled variables, large time delay and the harsh environment, it is critical to develop a framework for the key-quality parameters control in the cone thickener system (CTS). Howbeit, the current method cannot finish the high-quality criteria for the intelligent mining paste. Furthermore, the prediction and control are different to fully considered in a unified framework. To address these issues, this paper proposes a new unified optimal prediction control framework (ConFrame) for the industrial mining system. The main implementation is divided into two stages, 1) the prior prediction model is built for the key-quality forecasting in the time series, and 2) a complete prediction model is then used as the core objective function for accomplishing the optimal control framework and parameter solution. The proposed framework accomplishes the integration of optimal and nonlinear control for the industrial paste-filling system. In addition, the presented framework narrows the gap between the mining concentration prediction and optimal control and provides a unified consideration for the full-scale pasting-filling process. Furthermore, the approximation ability and convergence of the proposed algorithm are theoretically analyzed. Finally, the comparative performance is carried out on the numerical Rastrigin benchmark case to highlight the efficacy of the presented framework over the state-of-the-art framework. The presented framework is further tested in the practical cone thickener process which presents the prediction and control performance with high accuracy and robustness.