Modeling of chemical exergy of agricultural biomass using improved general regression neural network

• Published in: Energy (Oxford) 2016-11, Vol.114, p.1164-1175
• Main Authors: Huang, Y.W., Chen, M.Q., Li, Y., Guo, J.
• Format: Article
• Language: English
• Subjects: Artificial neural network model; Auto-refinement algorithm; Biomass; Chemical exergy; Element composition
• ISSN: 0360-5442
• DOI: 10.1016/j.energy.2016.08.090

A comprehensive evaluation for energy potential contained in agricultural biomass was a vital step for energy utilization of agricultural biomass. The chemical exergy of typical agricultural biomass was evaluated based on the second law of thermodynamics. The chemical exergy was significantly influenced by C and O elements rather than H element. The standard entropy of the samples also was examined based on their element compositions. Two predicted models of the chemical exergy were developed, which referred to a general regression neural network model based upon the element composition, and a linear model based upon the high heat value. An auto-refinement algorithm was firstly developed to improve the performance of regression neural network model. The developed general regression neural network model with K-fold cross-validation had a better ability for predicting the chemical exergy than the linear model, which had lower predicted errors (±1.5%). •Chemical exergies of agricultural biomass were evaluated based upon fifty samples.•Values for the standard entropy of agricultural biomass samples were calculated.•A linear relationship between chemical exergy and HHV of samples was detected.•An improved GRNN prediction model for the chemical exergy of biomass was developed.