Artificial Intelligence for Energy!

Summary:

Recently, Large Language Models (LLMs) have seen unprecedented growth, marked by significant advancements in foundation models, generative AI, reasoning capabilities, and intelligent agents. This progress has enabled innovations in battery conditioning monitoring, diagnostic and optimised control, enabling NVIDIA GPU AI acceleration of complex physical models using Physics-Informed Neural Networks (PINN). The presentation will demonstrate spatial-distribution of temperature and electrodes concentration AI accelerated prediction for LIB with a discussion on possible BMS implementation in order to improve the safety and sustainability of future battery system

Synthetic data generation…

Although digitalization of future power grids offers several coordination incentives, the reliability and security of information and communication technologies (ICT) hinders its overall performance. Inspired from the field of computational neuroscience, Spike Talk builds on a fine-grained parallelism on the information transfer theory in our brains, particularly when neurons (modeled as DERs) transmit information (inferred from power streams measurable at each DER) through synapses (modeled as tie-lines). Not only does Spike Talk simplify and address the current bottlenecks of the cyber-physical architecture by eliminating the ICT layer, it provides intrinsic operational and cost-effective opportunities in terms of infrastructure development, computations and modeling.

The third part of the tutorial will provide a pedagogic illustration of the key concepts and design theories by taking a simple example of Spike Talk’s operation principle in microgrids.

Motivation:

This tutorial opens a window in the future of energy management where explosive increase in EV and RE generation along with imperfect battery safety could limit the planned green energy transition. We are presenting three technologies: PINN and GAI along with a new type of Ai called Spike Talk that can works together in a holistic manner. The content will be presented at a not-advanced technical level, more conceptual, in order to be easier to understand for professionals wishing to move into AI for Energy realm.

Presenters:

Remus Teodorescu, Xin Sui and  Subham Sahoo
Aalborg University, Denmark

Target Audience:

All people with background or interest in computer science, mathematics, control or energy applications are welcome. The content will be presented at a not-advanced technical level, more conceptual, in order to be easier to understand for professionals wishing to move into AI for Energy realm.

Outline of Tutorial:

• Accelerated prediction with PINN in complex systems – Remus Teodorescu – 1 h
  1. PINN concept
  2. Application to SPM/DFN battery modelling
  3. Applciatiojn to thermal spatial distribution in batteries
• Improved state-estimation with synthetic data generation Xin Sui – 1h
  1. Introduction to GAI
  2. Data augmentation technology
  3. Synthetic generation with GAN for battery SOC estimation
• Spike Talk: Reflecting on a Sustainable AI Experiment for Power Electronics – Subham – 1h
  1. Introduction to neuromorphic AI
  2. Modeling anatomy between neurons and power electronics
  3. Data processing, continual online learning