The optimization of biodiesel production from waste cooking oil catalyzed by ostrich-eggshell derived CaO through various machine learning approaches

Abstract

The continuous increase in demand for fossil-based fuel has led to the requirement for an alternative source that must be renewable. Biodiesel is gaining global acceptance as a renewable source of energy. This research focuses on the optimization of the transesterification of waste cooking oil under the CaO-based catalyst derived from a solid ostrich eggshell by different types of machine learning approaches. The objective of the current study is to evaluate and compare the prediction results as well as the simulating efficiency of the biodiesel production yield using heterogeneous catalysts by various machine learning (ML) techniques: type 1 fuzzy logic system (T1FLS), response surface methodology (RSM), adaptive neuro-fuzzy inference system (ANFIS), and type 2 fuzzy inference logic system (T2FLS). The influence of the independent variables, methanol-oil molar ratio (M:O), temperature, catalyst concentration, and reaction time on the production yield was investigated. Among all the input parameters, the reaction temperature is the most influential one based on the aforesaid techniques. The validity of the proposed models has been verified with the help of statistical analysis and multiple linear regression. The values of the determination coefficient (R2) of type 2 fuzzy logic systems are 99.1% whereas R2 of type 1 fuzzy logic systems, response surface methodology, and adaptive neuro-fuzzy inference systems are 95.3%, 93.3%, and 83.2% respectively. All models give close predicted values. However, the type 2 fuzzy logic models were more accurate compared to other models. This proves that it is more capable of handling a wide range of dynamic processes in the chemical industry.