From Black Box to Clarity: AI-Powered Smart Grid Optimization with Kolmogorov-Arnold Networks

TL;DR

This paper introduces a novel AI approach using Kolmogorov-Arnold Networks for transparent and effective optimization of smart grids, addressing complex uncertainties in power flow management.

Contribution

It pioneers the application of Kolmogorov-Arnold Networks in smart grid optimization, providing a framework that enhances interpretability and handles uncertainties.

Findings

Effective optimization of hybrid AC/DC systems demonstrated

Improved interpretability of AI models in power systems shown

Framework successfully manages complex uncertainties in power flow

Abstract

This work is the first to adopt Kolmogorov-Arnold Networks (KAN), a recent breakthrough in artificial intelligence, for smart grid optimizations. To fully leverage KAN's interpretability, a general framework is proposed considering complex uncertainties. The stochastic optimal power flow problem in hybrid AC/DC systems is chosen as a particularly tough case study for demonstrating the effectiveness of this framework.