Gaussian Process-Based Economic MPC with Delay Compensation for SNCR-SCR Combined Denitration Systems

Selective noncatalytic reduction (SNCR) and selective catalytic reduction (SCR) denitration systems are the key units for reducing nitrogen oxide (NOx) emission from thermal power plants. Aiming at the NOx emission requirement and economic cost reduction, this paper proposes a Gaussian process-based economic model predictive control (MPC) approach with delay compensation for SNCR-SCR combined denitration systems. A combined denitration model is developed to describe the dynamics of the SNCR-SCR denitration process, which is then validated using industrial data. Subsequently, the Gaussian process model is introduced to predict the future NOx inlet concentration over the input delay period and prediction horizon. To cope with delays in the denitration systems, the predictor feedback delay compensation method is applied to compute state predictions after the input delay. Furthermore, the economic MPC method is designed to address both the NOx emission and the economic performance of the denitration systems. Finally, some comparative simulation results demonstrate the advantages of the proposed approach in terms of NOx emission compliance and economic optimization.