Identification of Compound in Wastewater using Artificial Neural Networks

Environmental pollution can be caused by wastewater disposal that is not handled properly, causing damage to groundwater, river and ocean ecosystems. Wastewater contains two types of compounds, namely organic and inorganic compounds. To design a wastewater treatment system, main and crucial step required is identifying pollutants in wastewater. The simplest identification of organic or inorganic molecules is to use spectroscopic techniques by evaluating their absorption spectrum for visible light. In this research, a discrete spectrum-based tool was designed to identify presence of organic and inorganic molecules in wastewater. This tool utilizes a discrete spectrum using 5 wavelengths, resulting in incomplete spectral information. For this reason, this discrete spectrum-based tool is equipped with an Artificial Neural Network (ANN) algorithm with a Feed Forward Neural Network type Backpropagation approach. Performance of this ANN model produces an accuracy value of 70% and RSME 600,098 at epoch 32 for molecular classification and 13 for concentration.