A Machine Learning Approach to Feature Selection and Uncertainty Analysis for Biogas Production in Wastewater Treatment Plants
The increasing demand for effective waste management and renewable energy has spurred research into predicting biogas production in wastewater treatment plants. This study presents a comprehensive process, beginning with data collection from an operational plant, followed by detailed analysis to address potential issues. A key contribution is the development of a robust machine learning model, fine-tuned using advanced optimization techniques. To enhance its practical utility, prediction intervals were integrated to quantify uncertainty, enabling regulators to make informed decisions. The model demonstrated strong performance, explaining 82% of the variability in the test data and delivering predictions closely aligned with actual biogas production. This reliability supports confident decision-making in wastewater treatment operations. The study also identified critical factors influencing biogas output, including sludge characteristics, operational practices, and sludge quantity. By focusing on these variables, operators can optimize processes and significantly improve biogas yields. This predictive capability, combined with an understanding of influencing factors and quantified reliability, offers significant advantages. It enables operators to enhance plant efficiency and biogas production while providing regulators with reliable predictions to guide policy and resource management. These advancements contribute to more sustainable and efficient waste management practices.
All source code used to generate the results is available in Jupyter Notebook file. All data and materials are available upon request from the corresponding author.
The following dependencies are required to run the code:
pandas==2.2.2
numpy==1.26.4
matplotlib==3.10.0
seaborn==0.13.2
scienceplots==2.1.1
scikit-learn==1.5.2
lightgbm==4.5.0
optuna==4.2.0
optuna-integration==4.2.1
scipy==1.13.1
tqdm==4.67.1
puncc==0.8.0
mapie==0.9.2
These dependencies are listed in the requirements.txt file. To install them, run:
pip install -r requirements.txtAll source code is made available under the BSD 3-Clause License. You are free to use and modify the code without warranty, provided that you provide attribution to the authors. For the full license text, see the LICENSE file.
- Mahsa Samkhaniani
- Shabnam Sadri Moghaddam
- Hassan Mesghali
- Amirhossein Ghajari
- Nima Gozalpour
Currently under review. Link will be added upon acceptance.