Efficient water quality prediction by synthesizing seven heavy metal parameters using deep neural network

Abstract

Water quality is primarily assessed by consumers based on its aesthetic acceptability, encompassing factors such as taste, odor, and appearance. The seven heavy metal parameters - magnesium, sulfate, potassium, sodium, calcium, total hardness, and chloride - significantly impact the aesthetic acceptability of water, playing crucial roles in influencing factors such as taste and odor. In this study, we present a deep learning-based approach to reconstruct these seven parameters from four parameters (temperature, electrical conductivity, potential of hydrogen, and total dissolved solids) in order to estimate the water quality index. This approach simplifies data collection, making it cost-effective without the need for expensive devices or traditional lab facilities. It also facilitates the assessment of sample quality by calculating the water quality index using these seven synthesized parameters in conjunction with recorded potential of hydrogen and total dissolved solids values. Accurate assessment of these parameters is essential for evaluating the suitability of water for various purposes, including consumption and industrial applications. Our proposed model is trained on a comprehensive dataset encompassing a wide range of water samples. Through extensive experimentation and evaluation, we demonstrate that our deep learning model effectively reconstructs the seven parameters with high accuracy and reliability. Our study's results demonstrate that utilizing reconstructed parameters achieves performance comparable to that of using actual recorded parameters for determining the quality of each sample. This approach offers a cost-effective and efficient solution for understanding water quality, enabling improved decision-making and management of water resources.