Wastewater Quality Indicator Estimation Using Machine Learning and Data Augmentation Techniques

We propose a novel design methodology for the estimation of the quality of wastewater using Ultraviolet Visible (UV-Vis) spectroscopy and Machine Learning. Addressing the challenge posed by limited real-world data, particularly in highly polluted industrial environments, this study introduces a data augmentation method based on Conditional Generative Adversarial Networks (CGAN). The effectiveness of this method is evaluated by creating a regression model based on a Multi-layer Perceptron (MLP) to estimate the chemical oxygen demand, a water quality indicator, using the UV-Vis absorption spectrum. The proposed method demonstrates that insufficient wastewater sample data can be augmented to improve the performance of the regression task for chemical oxygen demand (COD) estimation.