Ultrasound-Based Indirect Water Temperature Measurement Using Machine Learning

Maintaining water quality is a key environmental challenge, with temperature being a critical factor. real-time monitoring is essential as traditional methods are expensive and slow. Acoustic sensing, which uses sound waves to track temperature, turbidity, and pollutants, offers a promising solution. Enhanced by machine learning (ML), it enables pattern detection, autonomous classification, and predictive analysis. This paper explores a novel approach that integrates acoustic sensing and ML to enhance water quality monitoring. Controlled experiments were conducted to analyze the use of ultrasonic signals for estimating water temperature. Regression analysis revealed that the KNeighbors Regressor outperformed other models in predicting water temperature, achieving perfect accuracy with zero error based on the Mean Squared Error. Classification algorithms such as Random Forests and Support Vector Machines also revealed exceptional performance in distinguishing between hot (27-33°C) and cold (14-17°C) water conditions. The results highlight the potential of acoustic-based sensing, enhanced by ML techniques, in environmental monitoring, demonstrating its effectiveness for continuous assessment.