Abstract The rapid digitalization and decarbonization of electric power systems have created vast, high‐velocity data streams and complex decision problems. Machine learning (ML)—spanning supervised, unsupervised, reinforcement, and physics-informed paradigms—has emerged as a core toolkit for forecasting, monitoring, control, optimization, and cybersecurity across grid planning, operations, and markets. This review synthesizes recent advances with an application-oriented taxonomy: (i) forecasting and situational awareness; (ii) protection, fault diagnosis, and resilience; (iii) security assessment and state estimation; (iv) optimization and control (including learning-assisted optimal power flow); and (v) emerging graph-, physics-, and reinforcement-learning approaches. We discuss data and model lifecycle issues (labels, drift, uncertainty, explainability, and MLOps), benchmarking needs, and pathways for trustworthy deployment. We conclude with a research agenda for grid-aligned, auditable ML at scale. Keywords: Deep Learning, Forecasting, Graph Neural Networks, Machine Learning, Optimal Power Flow, Physics-Informed Learning, Power Systems, State Estimation, Reinforcement Learning.
Shajin et al. (Fri,) studied this question.