Advancing smart city sustainability with Internet of Things and artificial intelligence aided low-cost digital twin systems for waste management

This study explores the transformative potential of low-cost digital twin systems, integrated with IoT, AI, and machine learning, to enable real-time monitoring, efficient management, and predictive insights for municipal solid waste, offering an optimized solution for smart cities. A digital twin framework is proposed with possible use cases. To minimize costs, a LoRA-enabled basic bin monitoring unit (BBMU) has been developed on the ESP32 platform. This unit provides real-time data on waste levels, gas concentrations, and humidity at various collection points, along with their corresponding GPS coordinates and is strategically installed on selected bins for the purpose of sample data collection. Trained on a dataset from Polish cities, different machine learning models were tested to predict waste generation across the city; the best and most efficient ANN model was chosen for deployment. The ANN model’s R2 score of 0.944 shows a significant correlation between the projected and actual values. A Pearson correlation coefficient of 0.998 implies a strong linear link between variables. Willmott’s index of agreement, 0.987, validated the excellent correlation of the ANN model’s predictions with the observed data. The ANN performed better than other models, as seen by their lower root mean squared error (8130.269) and mean absolute error (7656.989). The study highlights the cost-effectiveness and efficiency of digital twin technology in improving waste management practices, contributing to the sustainability of smart cities.