ARMAX Model Identification and Model Predictive Control for Coke Oven Gas Final Cooling Tower

he final cooling of coke oven gas (COG) is a critical unit in coking production, where precise temperature control is essential for product quality and energy efficiency. However, the process is characterized by large inertia, long time delays, and strong disturbances, making traditional PID strategies insufficient for optimal performance. To address these challenges, this study proposes a Model Predictive Control (MPC) strategy based on a data-driven Auto-Regressive Moving Average with eXogenous inputs (ARMAX) model. Using real historical data from an industrial chemical plant, a gray-box ARMAX model is identified to accurately capture the system's dynamic behaviors. This model serves as the predictive core for the MPC controller, which regulates the cooling water flow rate through rolling horizon optimization. Closed-loop simulation results demonstrate that the proposed framework significantly outperforms traditional PID control in terms of setpoint tracking, disturbance rejection, and stability. This work provides a robust solution for the advanced control of complex thermal processes, contributing to the intelligent and energy-efficient transformation of the chemical industry.he final cooling of coke oven gas (COG) is a critical unit in coking production, where precise temperature control is essential for product quality and energy efficiency. However, the process is characterized by large inertia, long time delays, and strong disturbances, making traditional PID strategies insufficient for optimal performance. To address these challenges, this study proposes a Model Predictive Control (MPC) strategy based on a data-driven Auto-Regressive Moving Average with eXogenous inputs (ARMAX) model. Using real historical data from an industrial chemical plant, a gray-box ARMAX model is identified to accurately capture the system's dynamic behaviors. This model serves as the predictive core for the MPC controller, which regulates the cooling water flow rate through rolling horizon optimization. Closed-loop simulation results demonstrate that the proposed framework significantly outperforms traditional PID control in terms of setpoint tracking, disturbance rejection, and stability. This work provides a robust solution for the advanced control of complex thermal processes, contributing to the intelligent and energy-efficient transformation of the chemical industry.