Artificial Neural Network–Guided Optimization of Microwave-Assisted Biodiesel Production Using a Zn–Ce-Loaded Agro-Waste Activated Carbon Bifunctional Catalyst

Catalyst limits and the intricate, nonlinear interaction of operating parameters make it difficult to produce biodiesel from low-cost, high-free fatty acid feedstocks in an efficient and sustainable manner. By combining artificial intelligence-based process optimization with a bifunctional Zn–Ce-loaded activated carbon catalyst made from agricultural waste, this study seeks to increase the efficiency of biodiesel synthesis. Artificial neural network (ANN) models were developed and assessed using experimental data from 52 runs of the microwave-assisted production of biodiesel from waste cooking oil. The best feedforward neural network was combined with a genetic algorithm (GA) to estimate biodiesel yield and determine ideal operating parameters after a number of ANN topologies were investigated. With a correlation coefficient of roughly 0.996 and a mean square error of roughly 5.7 × 10⁻², the chosen ANN model with three hidden layers showed outstanding prediction ability. With a maximum anticipated biodiesel yield of 98.57%, ANN–GA optimization found ideal parameters of 0.3 weight percent catalyst loading, an oil-to-methanol ratio of 1:7, a reaction duration of 4.3 minutes, and a microwave power of 97.5%. These findings show that coupling ANN-GA optimization with a sustainable bifunctional catalyst offers a precise and economical method for enhancing microwave-assisted biodiesel production, providing a promising route for the effective use of feedstocks obtained from trash.