Benchmarking State Space Models, Transformers, and Recurrent Networks for US Grid Forecasting

2026-02-24

AI Summary

This study addresses the challenge faced by power system operators in selecting optimal deep learning models for load and energy forecasting under varying data conditions. For the first time, it systematically evaluates the multi-step forecasting performance of five neural architectures—PowerMamba, S-Mamba, iTransformer, PatchTST, and LSTM—across six major U.S. power grids within a unified framework. Fair comparison is ensured through modular integration of weather covariates and standardized temporal feature processing. The findings reveal no universally superior model: when using only historical load data, PatchTST and state-space models (e.g., PowerMamba, S-Mamba) perform best; however, with weather data incorporated, iTransformer achieves significant gains. Performance further varies across solar, wind, and electricity price forecasting tasks, offering a data-driven guideline for practical model selection.

Technology Category

• Machine Learning: Deep Neural Architectures and Foundation Models
• Search and Optimization: Learning to Search
• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Selecting the right deep learning model for power grid forecasting is challenging, as performance heavily depends on the data available to the operator. This paper presents a comprehensive benchmark of five modern neural architectures: two state space models (PowerMamba, S-Mamba), two Transformers (iTransformer, PatchTST), and a traditional LSTM. We evaluate these models on hourly electricity demand across six diverse US power grids for forecast windows between 24 and 168 hours. To ensure a fair comparison, we adapt each model with specialized temporal processing and a modular layer that cleanly integrates weather covariates. Our results reveal that there is no single best model for all situations. When forecasting using only historical load, PatchTST and the state space models provide the highest accuracy. However, when explicit weather data is added to the inputs, the rankings reverse: iTransformer improves its accuracy three times more efficiently than PatchTST. By controlling for model size, we confirm that this advantage stems from the architecture's inherent ability to mix information across different variables. Extending our evaluation to solar generation, wind power, and wholesale prices further demonstrates that model rankings depend on the forecast task: PatchTST excels on highly rhythmic signals like solar, while state space models are better suited for the chaotic fluctuations of wind and price. Ultimately, this benchmark provides grid operators with actionable guidelines for selecting the optimal forecasting architecture based on their specific data environments.

Problem

• power grid forecasting
• model selection
• deep learning architectures
• weather covariates
• forecasting tasks

Innovation

• State Space Models
• Transformers
• Electricity Forecasting
• Multivariate Time Series
• Model Benchmarking