Understanding the low-temperature drying process of sludge with machine learning in a sewage-source heat pump drying...

Area of Science:

• Environmental Engineering
• Chemical Engineering
• Data Science

Background:

• Sewage heat source heat pump drying is an emerging eco-friendly sludge treatment method.
• This technology reduces waste heat pollution and energy consumption.
• However, the specific drying characteristics of sludge using this method are not well understood.

Purpose of the Study:

• To develop and evaluate a novel sewage-source heat pump sludge low-temperature drying system.
• To model and analyze the sludge drying process using machine learning algorithms.
• To compare the performance of machine learning models against traditional numerical simulation models.

Main Methods:

• Construction of a novel sewage-source heat pump sludge low-temperature drying system.
• Application of machine learning algorithms, including XGBoost, combined with sludge drying theory.
• Bayesian optimization was employed to enhance model performance.

Main Results:

• Machine learning models significantly outperformed existing numerical simulation models.
• The XGBoost model, after Bayesian optimization, demonstrated superior prediction accuracy.
• Interpretable analysis revealed external air parameters (temperature, humidity) dominate the constant rate stage, while internal sludge characteristics (dry basis moisture content) become critical during the falling rate stage.

Conclusions:

• Machine learning models are feasible and effective for predicting and explaining low-temperature sludge drying processes.
• The study provides valuable insights for developing and managing sludge drying strategies using machine learning.
• This research highlights the potential of integrating data-driven approaches into environmental engineering applications.