The rapid growth of data traffic and the expansion of IoT devices have made energy consumption in communication networks a serious challenge to sustainable development. In this context, machine learning has emerged as an effective approach for optimizing energy-aware routing and resource management, playing a crucial role in intelligent networks. This survey article examines supervised, unsupervised, reinforcement, and deep learning approaches in areas such as wireless sensor networks, IoT, software-defined networks, and 5G/6G technologies, and presents a novel ML-based taxonomy for energy management.

The study further analyzes key performance indicators including energy efficiency, latency, throughput, and computational overhead, while addressing challenges such as the lack of standardized datasets, model complexity, and limited generalizability of results. This research demonstrates that integrating machine learning with sustainability principles can lead to the development of intelligent, low-power networks. Furthermore, future research directions are proposed for designing scalable, interpretable, and adaptive algorithms that align with the requirements of modern networks.