water quality detection system

Abstract

—Water quality is an important factor affecting human health and ecosystem sustainability; however, conventional laboratory testing requires time, cost, and skilled personnel, making it less efficient for routine monitoring. This study designs and develops an Internet of Things (IoT)-based water quality detection system that integrates pH, Dissolved Oxygen (DO), and turbidity sensors with an ESP32 microcontroller and applies the Support Vector Machine (SVM) method to automatically classify water conditions into good or polluted categories. Data from the three sensors are acquired in real time, normalized, formed into feature vectors, and then classified using the SVM model. The classification results are designed to be displayed through a local monitoring interface. Since hardware testing and model training have not yet been conducted, the results section is presented as a benchmark based on related studies. These studies indicate that SVM, particularly with the Radial Basis Function (RBF) kernel, can achieve classification accuracies ranging from 92% to 100% under controlled parameter conditions. Based on these findings, the proposed system targets a minimum classification accuracy of 90% and sensor reading errors below 5%. The system is expected to provide a fast, portable, and economical alternative for water quality monitoring as a complement to conventional laboratory testing methods.