Adaptive Nonlinear Model Predictive Control of NOx Emissions under Load Constraints in Power Plant Boilers

Abstract

Nitrogen oxide (NOx) emissions are major pollutants of coal-fired boilers. An adaptive nonlinear model-predictive control approach is presented to reduce NOx emissions of power plant boilers. Firstly, the boiler load and the NOx emissions are dynamically predicted by a differential evolution-based least-square support vector machine. Subsequently, based on data-driven prediction modeling, a nonlinear optimization model, with load and capacity constraints, is proposed for NOx emission minimization. Finally, a differential evolution algorithm is used to solve this optimization problem and obtain the optimal control variable settings. Experimental results based on practical data indicate that the proposed approach exhibits a promising performance in the prediction of the boiler load and NOx emissions. Compared with that obtained using the normal control strategy, the proposed approach can reduce NOx emissions by 3.2% and 4.3% under increasing and decreasing loads, respectively.