Optimizing Plastic Waste-to-Energy Systems: A Real-Time Monitoring and Prediction Framework Using IoT and Gradient Boosting

Abstract

Plastic waste-to-energy (WtE) systems hold immense potential to address the global challenge of plastic pollution while generating valuable energy resources. However, these systems cannot operate at optimal efficiency due to fluctuations in the composition of the feedstock, poor process conditions, and lack of proper monitoring in real-time resulting in fluctuations in energy generated and emissions. Modern solutions are nongenerative and do not have adaptation mechanisms, making them too dependent upon their configurations and cannot be easily scaled up or maintained for the long term. To tackle these problems, this research proposes combining IoT-based real-time data acquisition with sophisticated Gradient Boosting Machine Learning (ML) algorithms for real-time adjustments and predictions in WtE systems. TheIoT sensors gather different feedstock properties, reactor conditions, and emissions, providing data streams to a gradient-boosting-based prediction engine. This engine predicts basic and major parameters like gross energy ratio, emission, and efficiency of the waste conversion system. From these prediction data, a multi-objective optimization module adaptively controls the operation parameters to attain the maximum possible energy reuse and a minimum environmental influence. In the following, some benefits of the proposed solution are presented: better prediction error measure, real-time decision-making, and the potential to minimize GC emissions. The pilot implementation carried out in the MW-scale WtE plant showed up to 20% improvement in efficiency as well as decreased emissions in comparison with conventional systems. This study in particular timely contributes to the state-of-the-art of plastic WtE technologies by providing a scalable, data-validated, and green solution that idealistically connects the present deficiency between theoretical production and practical implementation.