Real-Time Cr(VI) Concentration Monitoring in Chrome Plating Wastewater Using RGB Sensor and Machine Learning

The transition to the 4th Industrial Revolution (4IR) in the electroplating industry necessitates intelligent, real-time monitoring systems to replace traditional, time-consuming offline analysis. In this study, we developed a cost-effective, automated measurement system for hexavalent chromium (Cr(VI)) in plating wastewater using an Arduino-based RGB sensor. Unlike conventional single-variable approaches, we conducted a comprehensive feature sensitivity analysis on multi-sensor data (including pH, ORP, and EC). While electrochemical sensors were found to be susceptible to pH interference, the analysis identified that the Red and Green optical channels are the most critical indicators due to the distinct chromatic characteristics of Cr(VI). Specifically, the combination of these two channels effectively functions as a dual-variable sensing mechanism, compensating for potential interferences. To optimize prediction accuracy, a systematic machine learning strategy was employed. While the Convolutional Neural Network (CNN) achieved the highest classification accuracy of 89% for initial screening, a polynomial regression algorithm was ultimately implemented to model the non-linear relationship between sensor outputs and concentration. The derived regression model achieved an excellent determination coefficient (R2 = 0.997), effectively compensating for optical saturation effects at high concentrations. Furthermore, by integrating this sensing model with the chemical stoichiometry of the reduction process, the proposed system enables the precise, automated dosing of reducing agents. This capability facilitates the establishment of a “Digital Twin” for wastewater treatment, offering a practical ICT (Information and Communication Technology)-based solution for autonomous process control and strict environmental compliance.