A Machine Learning Approach to Intelligent Streetlight Control for Sustainable Energy Management in Smart Cities

Abstract

This study examines how machine learning based intelligent streetlight control can improve sustainable energy management in smart cities by addressing the persistent problem of traditional fixed time streetlighting, which often consumes excessive electricity, increases municipal operating costs, and lacks real time responsiveness to traffic, pedestrian movement, ambient light, weather, and fault conditions. The purpose of the study is to evaluate the role of machine learning, IoT sensors, adaptive dimming, predictive brightness control, and remote monitoring in transforming conventional streetlights into intelligent urban energy assets. A quantitative, cross sectional, case-based research design was adopted using secondary evidence from cloud enabled, edge supported, and enterprise level smart city streetlighting cases reported in the literature. The sample consisted of selected smart lighting cases involving IoT connected streetlights, cloud management platforms, sensor-based control systems, and machine learning enabled prediction models. The key variables included machine learning prediction capability, IoT and sensor integration, adaptive dimming, predictive control, energy saving percentage, maintenance improvement, operational reliability, and sustainability outcomes. The analysis plan involved thematic coding, cross case comparison, Likert based evidence scoring, and numeric synthesis using the Energy Saving Percentage formula to compare traditional streetlighting with intelligent ML based systems. The findings indicate strong quantitative support for ML based intelligent lighting, with an overall support score of 4.42 out of 5.00, 86% of reviewed studies supporting ML or AI based streetlight control, IoT and sensor integration scoring 4.55, machine learning prediction and optimization scoring 4.48, and sustainability outcomes scoring 4.31. Energy saving patterns ranged from 25% to 60%, with an estimated average saving of 38.7%. Comparative results showed that traditional fixed time systems scored only 2.10, sensor based adaptive systems scored 3.95, and ML based predictive systems scored 4.62. The study implies that smart cities can reduce energy waste, improve lighting reliability, lower operational costs, support predictive maintenance, and strengthen sustainability goals by integrating machine learning with IoT based streetlight infrastructure.