Deep neural network-based multi-objective optimization of NOx emission and profit by recovering lignocellulosic biomass

In the pulp and paper industry, the external energy and pulping chemical consumption have been reduced by recovering the lignocellulosic biomass (LB) produced during the pulping process. However, this involves the inevitable emission of thermal NOx owing to the high pyrolysis reaction temperature. Therefore, it is necessary to simultaneously optimize energy and pulping chemical recovery and minimize NOx emissions. Hence, this study focuses on the multi-objective optimization of maximizing the net profit from energy and pulping chemical recovery while minimizing NOx emissions in recovering LB. For multi-objective optimization, a deep neural network (DNN)-based optimization model for the net profit and NOx emissions was developed with the 1,071 simulation data points according to the operating conditions. Consequently, Pareto-optimal solutions with profits between 5,241,520 and 1,329,558 $/y and NOx emissions between 87.95 and 78.27 ppm were obtained. The proposed Pareto-optimal front can offer comprehensive solutions to decision-makers.