Nonlinear black-box model for mesophilic anaerobic digestion

Abstract

This paper discusses developing a nonlinear black-box model technique to simulate the mesophilic anaerobic digestion (AD) process and predict output of biogas produced under various conditions. Black-box model technique utilises data generated by a system to define correlations between input and output variables of the system. It requires no in-depth knowledge of the system under consideration but the model is constructed by identifying these relationships to predict the information of the selected output. In the case of AD, input variables that influence the biogas production are reactor temperature, pH, mixing speed and pressure. The black-box model is constructed using MATLAB for analysing the input-output data from a bio gas plant. Two nonlinear model structures, nonlinear autoregressive exogenous model (NARX) with wavelet estimator and nonlinear Hammerstein-Wiener model (NLHW) with piecewise estimator are applied. Six model orders are chosen by trial and error are utilised to estimate the performance of the model. The performance of the models is compared with two different goodness-of-fit (GOF) measures. Finally, the model that best fits the experimental data is selected. The study found that the NLHW model structure with model order 230 is the model that best fits the experimental data. The model can be implemented in automated control systems of the biogas plant to optimise the plant operation in response to various process influences and improve the biogas production compared