A Multilayer Perceptron Feedforward Neural Network and Particle Swarm Optimization Algorithm for Optimizing Biogas Production

Efficient biogas production significantly impacts greenhouse gas (GHG) emissions and carbon sequestration by reducing emissions and enhancing carbon storage. Nonetheless, the consistency and optimization of biogas production are hindered by fluctuations in key input variables, namely, pH, moisture content, organic loading rate (OLR), and temperature, which significantly impact the quality of agricultural waste biomass and biogas production. Any fluctuations in these variables can affect biogas productivity. This study aims to provide valuable optimization parameters for maximum biogas production using rice straw and cow dung as agricultural waste biomass for biogas input materials. Therefore, machine learning techniques such as multilayer perceptron feedforward neural networks with a particle swarm optimization (PSO) combination generate optimal values for each variable for maximum biogas production. This study uses three variants of the training function for neural networks, namely gradient descent with momentum and adaptive learning rate, gradient descent with momentum, and gradient descent with adaptive learning rate. The findings reveal that, under an optimum pH value of 6.0000, a humidity of 62.3176%, an OLR of 67.6823 kg.m³/day, and a temperature of 37.0482 °C, biogas production has the potential to increase to 2.91 m³/day with a high accuracy testing value of R² = 0.90. These methods in use accurately predict the optimal parameters, with a maximum deviation of 8.48% from experimentally derived values and mean square error (MSE) of 0.0051243. This study emphasizes the benefits of using multilayer perceptron feedforward neural networks and particle swarm optimization to optimize operational parameters and accurately predict biogas production.