Regresar

Artificial neuronal networks to pbkp_redict the emissions of carbon dioxide (Co2) using a multilayer network with the levenberg-marquadt training method

Abstract:

This research work is based exclusively on the application of artificial neural networks, aimed at pbkp_redicting the CO2 pollution index. For the design of the ANN, a multilayer network of Backpropagation type has been created and the Levenberg-Marquardt method was used for its training. The neural network consists of three layers: input (Input), hidden (Hidden Layer) and output (Output); the architecture was generated with Matlab software. The model was validated with comparisons between real and forecasted values, with the interest of recognizing the trend of the index both in the short, medium and long term. Good quality results were obtained when the actual values and those pbkp_redicted by the system were checked, demonstrating that it is a highly accepted model for pbkp_rediction, favoring the planning processes.