A Data-driven Model for NOx Prediction in a Cement Clinker Calcination System Based on a Robust ESN

Nitrogen oxide (NOx) is one of the important atmospheric pollutants released during Cement material calcination process. Due to the complex physical and chemical reactions and harsh calcination environment, it is a challenging task to predict the NOx concentration in a cement clinker calcination system. To address this problem, this paper design a prediction model based on a robust echo state network. To improve the robustness to noise, Cauchy loss function and a mixed penalty term are introduced. Meanwhile, a corresponding iterative algorithm is proposed to optimize the parameters of the proposed model. Finally, experimental results on real data show the effectiveness of the proposed model.