Water quality monitoring system with arduino and k-nearest neighbors model

Ravi, Baraneeswar (2024) Water quality monitoring system with arduino and k-nearest neighbors model. Project Report. Melaka, Malaysia, Universiti Teknikal Malaysia Melaka. (Submitted)

Text (Full Text) Water quality monitoring system with arduino and k-nearest neighbors model.pdf - Submitted Version Download (1MB)

Abstract

The increasing demand for potable water necessitates advanced monitoring systems to ensure water quality. Monitoring water quality is one of the major objectives of this project where an ESP32 microcontroller which incorporates temperature, TDS, and pH Sensors is used. The first and foremost aim is to use machine learning for classifying the given water samples as either potable or not potable; this has been focused to avoid the issue of missing data values for the datasets and works to improve the prediction capability. Regarding the scope of the system, it identifies water samples by employing the k-Nearest Neighbors (kNN) algorithm the system uses real-time sensor reading stored in Firebase. Samples for data collection involve the measurements of various sensors that include temperature, Ph, and turblidity and makes sure we have a large data sample including those from potable and non-portable water sources. The preprocessing step involves cleaning to address cases of missing values and feature extraction to obtain the desirable attributes. Next, the preprocessed data is split into training and testing datasets, where the kNN model is trained with training data and employed to determine accuracy using the testing data. When the model has been proven to perform well in the simulation, it is then utilized in a real environment for water quality monitoring. It is important to have a clean and intuitive user interface of a mobile application where real-time data are shown and kNN will be used to make instant data retrieval with the help of Firebase solution. The system also includes a constant monitor and data acquisition of predefined sensors to detect water quality parameters that fall out of range. The system has been extensively tested and validated to confirm its measurability, with ongoing monitoring incorporated to handle issues that may appear. This paper’s objective lies in offering a sustainable and efficient solution for the periodical monitoring of water quality, facilitating solutions for providing people with safe drinking water. The dataset consists of 1,000 samples acquired from publicly available water quality datasets on the internet, ensuring a comprehensive range of potable and non-potable water scenarios. While real-time data collection is implemented using the ESP32 microcontroller and sensors, future development plans include expanding the use of IoT for continuous data acquisition and integrating a larger, more dynamic database. This will enable more robust real-time monitoring, as well as automatic alerts for potential water quality issues.

Actions (login required)

View Item