Characterization of suspended organic, inorganic, and microbial particles in water environment using an Electrical...

In aquatic environments, diverse particulate matter-including inorganic particles, organic particles, and microorganisms-significantly impacts water quality and ecological health. However, existing particle measurement technologies lack the capability to accurately distinguish and quantify these components in aquatic environments where particles are mixed. To address this, an integrated Electrical Sensing Zone (ESZ) and machine learning approach is developed, enhanced through systematic feature engineering. Two novel composite features-Time Product and Time Domain Product-are constructed to capture particle transit dynamics and waveform morphology, substantially improving separability between similar particle classes. The Support Vector Machine (SVM) model achieves an accuracy of 97.0%, with misclassification between organic particles and microorganisms reduced from 89.4% to 59.3% of total errors. Validation in mixed-particle systems demonstrates robust performance: concentration errors remain below 15.0%, and median particle size (D50) errors within 7.0%. The framework reliably distinguishes solid particles from bubble interference and offers practical advantages for field deployment. These results confirm the method's potential for real-time, in-situ water quality monitoring and multi-parameter particle characterization in complex aqueous environments.