Multi Model Remote Parameter Monitoring and Detection Using ML Algorithms for Pollution Prediction System

Abstract

Experimental work of Environmental pollution monitoring has become crucial for sustainable urban development and public health management. This paper presents a comprehensive real-time pollution monitoring system capable of simultaneously tracking multiple environmental parameters including air quality, carbon monoxide concentration, temperature, humidity, and dust. The system integrates an ESP32 microcontroller with multiple specialized sensors, GPS based geo-tagging capabilities, and cloud-based data visualization through Blynk IoT platform. A novel aspect of this work includes the implementation of machine learning algorithms for continuous predictive analytics, enabling forecasting of pollution trends based on historical data patterns. The system features local display capabilities via LCD screen and remote monitoring through IoT connectivity, making it suitable for both mobile and stationary deployment scenarios. Field deployment results demonstrate the system’s effectiveness in providing accurate, real-time environmental data with predictive capabilities that can assist in proactive pollution management strategies. The compact design, powered by a regulated DC supply with buck boost conversion, ensures portability and reliability for diverse environmental monitoring applications. The proposed work employs Random Forest algorithm with neural network LSTM and provides a accuracy of R² as 91% from different geotagged location along with validation parameters and the obtained is superior among conventional systems.