A novel method for total chlorine detection using machine learning with electrode arrays.

Abstract

Chlorine is a common natural water disinfectant, but it reacts with ammonia's nitrogen to form chloramines, which affects the accuracy of free chlorine measurement. In this case, total chlorine can be used as an indicator to evaluate the content of the effective disinfectant. In this article, a novel method to detect total chlorine using an electrode array in water has been proposed. We made the total chlorine sensor and captured the cyclic voltammetry curve of the electrode at different concentrations of chlorine ammonia. Principal component analysis and a peak sampling method were used to extract cyclic voltammetry curves, and the total chlorine prediction model was established by support the vector machine and extreme learning machine. The results show that the best predicting power was achieved by support vector regression with principal component analysis (R2 = 0.9689). This study provides a simple method for determining total chlorine under certain conditions and likely can be adapted to monitor disinfection and water treatment processes as well.