Fourier-Enhanced Recurrent Neural Networks for Electrical Load Time Series Downscaling
By: Qi Chen, Mihai Anitescu
Potential Business Impact:
Predicts electricity use more accurately.
We present a Fourier-enhanced recurrent neural network (RNN) for downscaling electrical loads. The model combines (i) a recurrent backbone driven by low-resolution inputs, (ii) explicit Fourier seasonal embeddings fused in latent space, and (iii) a self-attention layer that captures dependencies among high-resolution components within each period. Across four PJM territories, the approach yields RMSE lower and flatter horizon-wise than classical Prophet baselines (with and without seasonality/LAA) and than RNN ablations without attention or Fourier features.
Similar Papers
Recurrent Neural Networks with Linear Structures for Electricity Price Forecasting
Machine Learning (Stat)
Predicts electricity prices with better accuracy.
Reservoir Computing via Multi-Scale Random Fourier Features for Forecasting Fast-Slow Dynamical Systems
Neural and Evolutionary Computing
Predicts complex changes in nature and brains better.
Frequency-adaptive tensor neural networks for high-dimensional multi-scale problems
Machine Learning (CS)
Helps computers solve hard problems faster.