Development of biochar yield prediction model from food waste pyrolysis using explainable artificial intelligence algorithm

Esvaran, Surend (2024) Development of biochar yield prediction model from food waste pyrolysis using explainable artificial intelligence algorithm. Project Report. Melaka, Malaysia, Universiti Teknikal Malaysia Melaka. (Submitted)

Text (Full Text) Development of biochar yield prediction model from food waste pyrolysis using explainable artificial intelligence algorithm.pdf - Submitted Version Download (5MB)

Abstract

The collection of waste in the natural environment has increased to such an extent that new and viable solutions for food waste management are required. Biochar is a carbonaceous material produced by means of the pyrolysis process from biomass. It has received much attention due to its huge potential for use as a sustainable soil amendment, carbon sequestration practice, and also as a renewable energy resource. Food waste, one of the gigantic environmental problems, can be converted into biochar through the process of pyrolysis. Process conditions such as temperature and residence time along with feedstock characteristics do have an impact on the biochar yield from the pyrolysis of food wastes. Accurate prediction of biochar yield helps in effective food waste management and minimization of environmental impact through resource use optimization. Hence, AI algorithms have been tested for modeling complex systems with a view to predicting their outputs. This is in the domain of developing an AI-driven, accurate model for biochar yield prediction from the pyrolysis of food waste using Explainable AI techniques. Data collection, preprocessing, and feature selection will be done from the experiments on pyrolysis of food waste. Next, apply machine and deep learning techniques; models such as linear regression, random forest, K-nearest neighbors, and convolutional neural networks will be evaluated. Model transparency and interpretability will be attained using XAI methods, such as SHapley Additive exPlanations(SHAP) values and Local Interpretable Model-agnostic Explanation(LIME). It is expected that the expected deliverables of this research would be accurate AI-based prediction models; insight into the factors influencing biochar yield and process control will also be provided. Recommendations for the optimization of food waste pyrolysis processes to drive the transition toward sustainable biochar production will be derived. It will help cut down greenhouse gas emissions, improve agricultural productivity to enhance food security.

Actions (login required)

View Item