Machine Learning-Based Prediction of Biogas Production from Sludge Characteristics in Four Anaerobic Digesters: Development of the AD2Biogas Prediction Tool

One of the alternative ways to obtain low-cost energy is to use biogas generated by the digestion process from sewage sludge. This paper presents an analysis of the processes in four anaerobic digesters (AD) - A, B, C and D. The study analyzed the amount of biogas produced in each digester tank and compared them with each other. Using data sets consisting of parameters relating to the pre-sludge and surplus sludge diverted to each tank, the effect of the proportion of these parameters on biogas production efficiency was studied. Based on this data, several models using different machine learning techniques were built and compared, which can be used to support the biogas production optimization process. A free convenient web tool written in Python language - AD2Biogas Predictor Tool - was also given away for sewage treatment plants to conveniently estimate the predicted amount of biogas produced on a given day using the implemented models. The main objective of the study is to understand how the studied parameters affect the efficiency of the process and identify potential optimization strategies, as well as to propose a model for biogas yield prediction based on sludge characteristics. The result of the study is to contribute to increasing the efficiency of sludge management in wastewater treatment plants and increasing biogas production, both in the form of developed models and a software tool.

---

Opracowanie rekordu ze środków MNiSW, umowa nr POPUL/SP/0154/2024/02 w ramach programu "Społeczna odpowiedzialność nauki II" - moduł: Popularyzacja nauki (2025).