A new hybrid modeling and optimization algorithm for improving carbon efficiency based on different time scales in sintering process. (October 2019)

• Record Type: Journal Article
• Authors: Zhou, Kailong; Chen, Xin; Wu, Min; Cao, Weihua; Hu, Jie
• Is Part Of: Control engineering practice. Volume 91(2019)
• Publication Date: 2019-10
• Subjects: Sintering process -- Carbon efficiency -- Multiple time-scale -- Comprehensive carbon ratio (CCR) -- CO/CO2 -- Linear combination scheme with multiple NNs -- Multi-objective optimization; Automatic control -- Periodicals; 629.89
• DOI: 10.1016/j.conengprac.2019.104104
• Languages: English
• ISSNs: 0967-0661

Abstract

Abstract: Sintering process is a major energy-consumption process in steel making processes. Carbon efficiency, which reflects the level of the energy-consumption, is affected by the raw material variables and the operating variables. This situation makes the process with special dynamic characteristics, especially with different operation periods. However, there is seldom research on carbon efficiency with the multiple time-scale property taken into account. This paper introduces an optimization method for carbon efficiency based on an intelligent multiple time-scale model, which is able to optimize process variables in both long and short time scales. As the comprehensive carbon ratio (CCR) and the ratio between CO and CO 2 in exhaust gas (CO/CO 2 ) are taken as the carbon efficiency indexes, a predictive model consisting of two sub-models is developed, one for predicting the state variables with a single neural network (NN), and the other for predicting carbon efficiency indexes with a linear combination of NNs. Then a multi-objective, multi-time-scale optimization framework is designed, which is able to optimize carbon efficiency in two time scales, according to the optimization variables available encountered at different operation periods. Finally, the experimental results based on actual process data shows its feasibility and improvement in carbon efficiency optimization.