A Real-Time ETP Outlet Monitoring Framework Leveraging Environmental IoT, Colorimetry, and Learning Theory

Abstract

Industries such as textiles and ready-made garments (RMG) contribute significantly to industrial waste and effluent, leading to severe environmental pollution. Effluent treatment plants (ETPs) are designed to treat and recycle this wastewater, aiming to remove suspended solids before discharge. However, many factories worldwide either operate without ETPs or fail to maintain them consistently due to high energy consumption, resulting in the release of untreated waste and posing serious environmental risks. This research aims to establish a system for real-time monitoring of water quality outside industrial sites, enabling the determination of ETP operational status and providing an automated monitoring solution. This research utilizes IoT (Internet of Things)–based real-time monitoring to enhance industrial wastewater management. A Convolutional Neural Network (CNN) is used for video-based binary classification to detect the On/Off state of ETPs. K-Nearest Neighbor (KNN) classifies water quality based on sensor data, and Long Short-Term Memory (LSTM) forecasts seasonal impacts on water quality, including inactive periods. Additionally, colorimetry analysis is employed to monitor watercolor variations as potential indicators of contamination. The proposed system achieves accuracy of 98.3%, 97%, and 94.9% for CNN, KNN, and LSTM, respectively, offering a practical, low-cost solution for remote ETP monitoring. Furthermore, it holds promise for broader applications in environmental compliance, with potential benefits for the agriculture and nutrition sectors by minimizing industrial pollution and supporting sustainability efforts.

Keywords

• Effluent treatment plants
• E-IoT
• deep learning
• RGB colorimetry analysis
• video classification
• water quality index