Artificial Intelligence Techniques in Distribution Systems

Advancements made in artificial intelligence (AI) and machine learning have brought novel occasions to various complicated control and operational tasks in the systems of modern electric distribution. However, the applications of AI for different networks such as systems of cyber-physical face numerous challenges, including higher needs for the quantity and quality of training data, privacy issues, physical inconsistency, interpretability, and efficiency of data. The current work provides a systematic overview of the latest AI procedures emerging in the post-pandemic era. It explores advancements in graph learning, transfer learning, and the attention mechanism of deep learning. Furthermore, the study delves into the integration of these AI methodologies into physics-guided neural networks and reinforcement learning. This dual focus aims to elucidate the evolving landscape of artificial intelligence applications in the context of contemporary challenges and opportunities. A detailed categorization and investigation of recent research endeavors to harness such developments, such as power flow, voltage control, state estimation, line parameter calibration, and topology identification have been presented. The current chapter focuses on the integration of distributed energy resources, features of distribution system operation, and illuminating prospects, as well as challenges characterized by the privacy, interpretability, and explainability of these AI applications within smart distribution systems. Ultimately, an attempt is made to provide deeper insights into the concept of smart distribution systems by their interoperation with transportation electrification and smart building.