Distinguished seminar «Toward AI-Native Energy Systems»

Abstract

The conference will start with a few concrete, successful examples of AI in upcoming 6G networks, where learning-based methods already support prediction, optimization, and decision-making under uncertainty and strict constraints. Building on these lessons, Khormuji will draw an analogy to energy systems, which face the same core challenges: complex dynamics, partial observability, non-stationary environments, and the need for safety and reliability by design.
Then, he will outline a practical roadmap toward AI-native energy systems, combining (i) physics-guided forecasting and state estimation (e.g., spatio-temporal models and PINNs), (ii) continuously calibrated digital twins, and (iii) safe decision-making via constrained MPC and safe RL. As a flagship example, he will discuss wave-energy systems, which are modern yet highly dynamic and uncertainty-driven—an excellent “channeling” case where forecasting, digital-twin rollouts, and constrained control must work together to maximize power capture while respecting loads and operational limits.

Bio

Dr Nasiri Khormuji got a PhD in Telecommunications Engineering from the KTH-Royal Institute of Technology of Sweden. He was Visiting Research Scholar at Stanford University. Back to the KTH-Royal Institute of Technology he got a postdoctoral researcher position to conduct advanced research in information theory and wireless communication technologies. In 2013 he joined Huawei Sweden, first as Senior researcher and later as Principal Researcher. There he contributed to the development of 5G/6G technologies, led multidisciplinary teams, and shaped long-term research strategies. His background combines fundamental research (information theory, optimization, machine learning) with applied innovation in complex systems. These efforts relied heavily on AI methods for prediction, optimization, and intelligent control. A significant part of his work involved energy-efficient algorithm and system design, reflected across more than 60 scientific articles in IEEE flagship journals and conferences, and 60+ patent applications.