Modeling and Simulation of Alternative Energy Generation Processes using HONN

Abstract

This chapter deals with the application of Higher Order Neural Networks (HONN) on the modeling and simulation of two processes commonly used to produce gas with energy potential: anaerobic digestion and gasification. Two control strategies for anaerobic digestion are proposed in order to obtain high biomethane flow rate from degradation of organic wastes such as wastewater. A neurofuzzy scheme which is composed by a neural observer, a fuzzy supervisor, and two control actions is presented first. After that, a speed-gradient inverse optimal neural control for trajectory tracking is designed and applied to an anaerobic digestion model. The control law calculates dilution rate and bicarbonate in order to track a methane production reference trajectory under controlled conditions and avoid washout. A nonlinear discrete-time neural observer (RHONO) for unknown nonlinear systems in presence of external disturbances and parameter uncertainties is used to estimate the biomass concentration, substrate degradation, and inorganic carbon. On the other side, a high order neural network structure is developed for the process identification in a gasification reactor; the gas, composed mainly of hydrogen and carbon monoxide (synthesis gas or syngas), is produced from thermo chemical transformation of solid organic wastes. The identifier is developed in order to reproduce a kinetic model of a biomass gasifier. In both cases (biological and thermo chemical processes), the Extended Kalman Filter (EKF) is used as a training algorithm. The proposed methodologies application is illustrated via numerical simulations.