Real-time dynamic prediction model of carbon efficiency with working condition identification in sintering process. (March 2022)

• Record Type: Journal Article
• Authors: Hu, Jie; Wu, Min; Chen, Luefeng; Cao, Weihua; Pedrycz, Witold
• Is Part Of: Journal of process control. Volume 111(2022)
• Page Start: 97
• Page End: 105
• Publication Date: 2022-03
• Subjects: Iron ore sintering process -- Carbon efficiency prediction -- Real-time dynamic prediction model -- Working condition identification -- Broad learning; Process control -- Periodicals; Fabrication -- Contrôle -- Périodiques; Process control; Periodicals; Electronic journals; 660.281
• DOI: 10.1016/j.jprocont.2022.02.002
• Languages: English
• ISSNs: 0959-1524

Abstract

Abstract: Accurate prediction of carbon efficiency is a prerequisite for achieving energy saving and consumption reduction in an iron ore sintering process, and is the key to guaranteeing the quality and yield of sintered ore. This paper proposes an original real-time dynamic prediction model for carbon efficiency prediction in the process. A Savitzky–Golay filter is used to eliminate noise of the actual production data collected from a cooperative sintering plant, and the correlation between carbon efficiency and process parameters is determined by mutual information. A modified version of maximum entropy clustering algorithm is presented for identifying working conditions to accurately discriminate between anomalies and normal working conditions. Then, the real-time dynamic prediction model of carbon efficiency based on broad learning is established by taking into account the process characteristics and using the prediction error information under normal working conditions. The proposed model is demonstrated to be valid by carrying out some experiments with actual production data. The experimental comparative analysis show that this model has good generalization capabilities and high real-time prediction accuracy, and is superior to other advanced methods in dynamic prediction of carbon efficiency. Highlights: An improved maximum entropy clustering is presented to identify working conditions. A real-time dynamic prediction model of CCR based on broad learning is established. A suitable modeling framework is proposed to reflect the sintering process dynamics.