A Method of Prediction and Control of NOx Emission Reduction in Pure Oxygen Combustion Reheating Furnace

Abstract

When the traditional double-cross-limiting control method is used to control the furnace temperature of the full-oxygen heating furnace, the strong oxidizing nature of pure oxygen often causes the emissions of nitrogen oxides (NOx) to exceed the standard. In order to solve this problem, a new inverse double-cross-limiting predictive control method is proposed in this article. First, the reverse double-cross-limiting control method sets itself apart from the traditional one by initially opening the pure oxygen valve followed by the natural gas valve. In this way, by curbing excessive oxygen levels, the issue of NOx emissions surpassing regulatory limits can be effectively addressed. Second, based on bidirectional long and short memory neural network (BiLSTM) and Sparrow Search Algorithm (SSA) optimization, a new SSA-BiLSTM prediction model is proposed to monitor and predict the change trend of NOx in real time. Compared with other five prediction models, the SSA-BiLSTM prediction model has advantages in determination coefficient, root mean square error, and other evaluation indicators. Finally, the experimental results show that the SSA-BiLSTM prediction and inverse double-cross-limiting control method can effectively control the NOx emission below 40 mg m⁻³ while ensuring the normal temperature rise of the heating furnace.