Industrial-Scale Mathematical Modeling and Real-Time Predictive Software for Biogas Yield Optimization Using Regression Analysis and Big Data

Source abstract

This research examines computer and mathematical modeling of anaerobic digestion in livestock waste recycling. Various modeling approaches, including ADM1, are analyzed, and the devel-opment of a proprietary computer program based on a self-developed mathematical model is described. The proposed mathematical model of anaerobic fermentation has demonstrated effi-ciency and reliability under industrial-scale conditions. It was developed using regression analysis and correlation methods, establishing empirical relationships between biogas yield and 20, 26 key process parameters continuously measured by biosensors in real production settings. Despite a limited static dataset (only five measurement points), the model exhibited a low average ap-proximation error of 5.5% and a high determination coefficient of 0.95, confirming its accuracy and high quality. The computer program is designed for online statistical data transmission via cellular communication and the Internet, enabling the processing of large volumes of data in ac-cordance with Big Data concepts. Its architecture ensures flexibility, scalability, and precision. Furthermore, the integration of machine learning technologies is proposed to enhance the pro-gram's predictive capabilities, leveraging Big Data analytics. The proposed computer–mathematical model is thus considered a practical, accurate, and scalable solution for monitor-ing and optimizing industrial-scale anaerobic digestion processes, combining empirical modeling with modern data processing technologies.