Using Neural Network to Predict Pyrolysis Biochar Production and Its Impact on Low-Carbon Agricultural Development in Scotland

Sustainable agricultural development is one of the major challenges that call for effective actions. Agricultural residues have been long used as an important bioresource for energy and material recovery. In recent, significant attention has been paid to the production of biochar from the pyrolysis of agricultural waste, and the application biochar for soil amendment and carbon abatement.

Based on an extensive review of existing data, this study developed a framework consisting of artificial neural networks (ANN) and the least squares support vector machine (LS-SVM) modelling to predict the biochar yield and quality from the pyrolysis of agricultural waste biomass and the impacts of biochar soil application on crop productivity. The framework utilised the adaptive neuro-fuzzy inference system approach and is able to predict biochar production under different pyrolysis conditions such as feedstock types (compositions), temperature and heating rates. The framework was used to evaluate the carbon abatement and agriculture productivity enhancement potential of soil amendment in Scotland by biochar produced from major Scottish agricultural residues