Energy Savings in Public Lighting by Using Adaptive Street Lighting—A Framework for Energy-Savings Assessment and Machine Learning-Based Evaluation

This study examines energy savings and efficient use through the application of smart adaptive lighting within the framework of “smart energy”, “smart city”, and “smart village”. Adaptive street lighting systems have emerged as an effective solution for reducing energy consumption while maintaining traffic safety. However, existing studies typically evaluate energy performance under predefined traffic conditions or focus primarily on AI-based control strategies without systematically investigating the influence of object speed on energy savings. This study proposes a comprehensive methodology that combines CupCarbon traffic simulation, Shape-Preserving Cubic Hermite Interpolation (PCHIP), Monte Carlo simulation, sensitivity analysis, and machine learning to evaluate and predict the energy-saving performance of adaptive street lighting. Unlike previous approaches, the proposed framework establishes a continuous relationship between object speed and energy consumption, enabling the estimation of energy savings across the entire operating speed range while quantifying the effects of object speed, pole spacing, and pre-activation time. The machine learning models are employed to predict energy-saving results generated by the simulation framework. Six regression models were trained and validated using simulated datasets, with Gradient Boosting achieving the highest predictive accuracy. Moreover, the analysis demonstrated that adaptive street lighting scenarios operating at 50% power (50 W) under no-object conditions and 100% power (100 W) during object detection achieved energy savings of 19–37% per luminaire compared with conventional street lighting, depending on object speed. Furthermore, adaptive lighting operating at a constant 50% power (50 W) during object detection yielded substantially higher energy savings of 59–77% per luminaire, highlighting the significant influence of the lighting control strategy on overall energy efficiency. The results demonstrate that lower object speeds yield the greatest savings. The proposed methodology provides a robust and scalable framework for the design, optimization, and intelligent control of adaptive street lighting systems and offers a benchmark for evaluating the maximum theoretical energy-saving potential under controlled traffic conditions.