H_MLP: a Time Series Prediction Model for the Control of SCR Denitration Systems

The selective catalytic reduction (SCR) denitration system is of great significance in the field of environmental protection. However, due to the complexity of the reaction and the influence of multiple factors, it is challenging to achieve precise control. This study proposes an H_MLP neural network model based on the multilayer perceptron (MLP). It innovatively classifies and processes the input feature channels and introduces the predictive control method as an intermediate layer constraint. This constructs a model with high interpretability, decoupling degree, accuracy, and training efficiency. Additionally, a recursive greedy algorithm is designed to apply the model to system control. Simulation experiments show that this model outperforms traditional algorithms in stability and accuracy, providing a new and effective solution for SCR denitration system control.