Abstract: Chemical oxygen demand (COD) is an important indicator for measuring water pollution, however, traditional chemical detection methods suffer from complicated operations, long durations, and secondary pollution. This paper presents a new COD measurement method based on a dual-light source sensor. The method involves preparing water samples in the laboratory, using dual ultraviolet light sources of 255 nm and 265 nm for water quality detection, and predicting COD values using a random forest (RF) regression model. Experimental results show that the coefficient of determination R2 of the test set for the RF regression model is 0.9910, and the root mean square error (ERMS) is only 0.0804, demonstrating the superior performance of the model in COD prediction. Compared to traditional spectrometers, the designed dual-light source sensor significantly achieves lightweight, reduces hardware costs and enhances the real-time performance of COD detection, providing efficient technical support for online water quality monitoring.

Key words: UV-Vis spectroscopy, chemical oxygen demand, Lambert-Beer law, random forest