Eventos Anais de eventos
ENCIT 2022
19th Brazilian Congress of Thermal Sciences and Engineering
Application of artificial neural networks to predict NO emissions into biomasses combustion processes
Submission Author:
thalyssa monteiro , MA
Co-Authors:
thalyssa monteiro, Glauber Cruz
Presenter: thalyssa monteiro
doi://10.26678/ABCM.ENCIT2022.CIT22-0691
Abstract
Replacing part of the use of fossil fuels in energy generation is an attractive solution to reduce its harmful effects on the Earth's atmosphere. During fossil fuel burning, some air pollutants are generated, strongly aggravating the release of greenhouse gases. On the other hand, the use of biofuels reduces the rates that effetely contribute to the Planet's environmental degradation. Several numerical models are widely used as a promising alternative to predict the pollutants emissions from burning biomasses and are considered reliable, saving time and financial resources. This study applied artificial neural networks (ANNs) for the prediction of nitrogen oxide emissions (NO) under different biomasses combustion, using a database with 40 lignocellulosic materials. For evaluation of numerical prediction models, 6 precision criteria were established: Mean Square Error (MSE), Root Mean Square Error (RMSE), Average Absolute Error (AAE), Average Bias Error (ABE), Mean Absolute Error (MAE), and linear regression coefficient (R). It was observed that the feedforward backpropagation model with 10 hidden neurons on the first layer and 15 neurons on the second (FF10x15 model) obtained a precision of 99.98% to estimated and predicted values for the NO emissions of the different samples evaluated, i.e., 1639.14 and 1639.10 mg Nm-3, respectively, showing the best performance for the six criteria established. Results showed that numerical prediction presented an excellent performance, and great accuracy, and can serve as a viable alternative for obtaining the NO emissions from different biomasses combustion when the employ of certain experimental processes becomes more difficult.
Keywords
lignocellulosic biomass, air pollutants, numerical model, nitrogen oxides
