Abstract
Factors such as population growth, industrialization, and the widespread application of fertilizers and pesticides have had significant repercussions on water quality. Consequently, the development of models for predicting water quality proves highly advantageous for monitoring water contamination.
Water quality monitoring encompasses various aspects, including biological, chemical, and physical attributes, as well as its suitability for specific applications. Effective river basin management relies on critical tasks such as assessing surface water quality and analysing water quality parameters. These actions are crucial for implementing measures to reduce pollution and maintain acceptable tolerance levels.
In this work, a water purification system was designed on Autodesk Inventor. The system designed uses sand and carbon filtration and a reverse osmosis process for water treatment to meet the SANS (The South African National Standard) 241 standards. The SANS standard outlines the specifications to be met for water to be considered as safe for consumption by human beings. To aid in the development of the water purification system, a literature study was conducted to look at important aspects such as important components of a water purification system. Secondly, ANN was used to predict the water quality index. The focus was on three (3) ANN algorithms namely Bayesian Regularization, Scaled Conjugate Gradient, and the Levenberg-Marquardt. The algorithms were chosen since they can handle complex, nonlinear data relationships, ability to perform well when given new data, and has high computational efficiency when dealing with large water quality dataset. The result obtained proved that the Levenberg-Marquardt algorithm was the most effective. Compared to two other methods, Bayesian Regularization and Scaled Conjugate Gradient, the Levenberg-Marquardt algorithm showed little differences between its predictions and the actual results. The biggest gap observed was about 1.37%, and most differences were below 1%. In conclusion, the objectives for the research were achieved, the work presented in this research involves assessing the effectiveness of a water purification system to determine its optimal performance in treating contaminated water and the.
Keywords—Artificial Intelligence, Artificial Neural Network, Water quality index, Water purification