Neural Identification of Thermochemical Processes for Solid Wastes Transformation

摘要

This paper presents a neural network application to identify the behavior of the model for two thermochemical processes, which are used to transform organic solid wastes. The first model corresponds to the char reduction zone of a gasification process, including inputs signals. The second one corresponds to a fluidized bed sludge combustor focused on the dynamics of NOx formation. The identification presented in this work is based on a discrete-time recurrent high order neural network (RHONN), which is trained with an extended Kalman filter (EKF) algorithm. The objective is to reproduce with neural networks the different gaseous components production and to study different chemical reactions taking place inside the reactors; this allows analyzing theoretically the process behavior in order to better understand some of the main phenomena before a future experimental stage. The neural identifier is implemented in specialized software and its performance is illustrated via simulations considering several operating conditions.