Abstract
Due to the exponential growth of the global population, it is necessary to ensure a viable
framework for the accessibility of clean and safe water since water is one of the necessities of life.
The typical method for collecting water characteristics is to collect samples and forward them to a
research institute for discovery and examination. However, this strategy consumes much time and
lots of human resources. This study therefore assembles, tests, and assesses the usefulness of
hybrid Machine Learning (ML) and embedded Internet of Things (IoT) quality of surface and
groundwater stations.
This study aims to develop an IoT and Artificial Intelligence design to provide an online drinking
water quality and decision platform. It also aims to develop the system integrated with sensors to
make safe and unsafe water quality decisions using ML deployed into an IoT server for water
quality monitoring. Finally, this study aims to construct a prototype of the IoT and AI system and
its performance was evaluated using classifier and reliability matrices. These aimed to create and
integrate IoT and AI in effective water quality monitoring systems.
This study considered water's physical and chemical properties, including pH, TDS temperature,
turbidity, Dissolved Oxygen, Total Dissolved Solid (TDS), Oxidation and Reduction Potential
(ORP), and Electrical Conductivity (EC), to determine the degree of water contaminant presence
in drinking water. These properties are measured with smart sensors incorporated into the
ThinkSpeak web server. The data received from the sensors was analyzed with Support Vector
Machine (SVM) and Artificial Neural Network (ANN) ML models to predict the impurity level of
the water samples put to the test. A water treatment method was also added to provide an automated
corrective measure based on a certain water contaminant level.
The results obtained from this system showed that the ML and Embedded IoT-based water quality
monitoring systems could provide remote monitoring of water quality parameters
The study concluded that ANN ML models is more effective to SVM models to remotely monitor
safe and unsafe water conditions.
Keywords: Internet of Things; Machine Learning; Water quality monitoring, Remote Monitoring.