Hybrid modeling and online optimization strategy for improving carbon efficiency in iron ore sintering process

摘要

Iron ore sintering is the second most energy consuming process in ironmarking, and the main energy consumption comes from the combustion of carbon. In order to improve the utilization of carbon in the process, taking a comprehensive coke ratio (CCR) as a metric of carbon efficiency, we develop a hybrid prediction model for CCR. On this basis, we present an online optimization strategy. Through the analysis of the sintering mechanism, the key sintering parameters affecting CCR are determined by a Pearson's correlation coefficient analysis. Then, multiple operating conditions of the process are identified by a Fuzzy C-Means clustering. Multiple models for different operating conditions are built by least squares support vector machine, and combined by using a Takagi-Sugeno fusion scheme (i.e., T-S rule-based model) so that the CCR prediction model is formed. To achieve CCR optimization based on this model, an online optimization strategy based on a chaos particle swarm optimization is presented. The simulation involving actual run data verifies that the proposed modeling method exhibits high prediction accuracy. Moreover, the results of actual runs show that the proposed optimization method satisfies the requirements of the actual sintering production, where the CCR is reduced by 1.327 kg/t (on average). (C) 2019 Elsevier Inc. All rights reserved.