Smart Biodigester System for Bioenergy Production Using Artificial Neural Network

Abstract

This study proposes an Internet of Things (IoT)-based smart biodigester system integrated with an Artificial Neural Network (ANN) for methane gas prediction and automatic control in household-scale biogas production. Conventional household biodigesters generally lack real-time monitoring and intelligent control mechanisms, resulting in unstable methane quality and inefficient gas utilization. The proposed system employs an ESP32 microcontroller and multiple sensors to monitor fermentation parameters, including temperature, gas pressure, pH value, substrate volume, and methane concentration. Sensor data collected during a 29-day anaerobic fermentation process were transmitted to a mobile-based monitoring platform and processed using an ANN model to predict methane gas concentration. A total of 870 valid data samples were used for model training and evaluation. To improve gas quality, the produced biogas was purified using a three-stage filtration system consisting of water, steel wool, and silica gel filters. The ANN prediction output was integrated with a servo-based automatic control mechanism to regulate methane gas flow into the storage tank according to predefined quality thresholds. Experimental results demonstrate that the proposed ANN model achieved a Mean Absolute Error (MAE) of 27.8 ppm, Root Mean Square Error (RMSE) of 34.6 ppm, Mean Absolute Percentage Error (MAPE) of 3.21%, and coefficient of determination (R²) of 0.94, indicating high prediction accuracy and strong agreement with actual methane measurements. The integration of IoT monitoring, ANN-based prediction, and automatic gas flow control improves the safety, efficiency, and reliability of household-scale biogas systems.