Energy Forecasting and Time Series Prediction

The field of energy forecasting and time series prediction is moving towards the development of more accurate and robust models that can handle complex and dynamic systems. Recent research has focused on leveraging advanced deep learning techniques, such as LSTM networks and knowledge distillation, to improve prediction accuracy and adapt to changing conditions. The incorporation of additional features, such as weather conditions and energy generation mix, has also been shown to enhance model performance. Furthermore, innovative approaches, such as adaptive masking loss and expert fusion, are being explored to address the challenges of predicting extreme events and handling imbalanced data. Noteworthy papers include:

  • A study on adaptive online learning with LSTM networks for energy price prediction, which introduced a novel custom loss function and online learning approach to improve prediction accuracy.
  • A paper on xTime, a framework for extreme event prediction that leverages hierarchical knowledge distillation and expert fusion to improve forecasting performance on rare events.

Sources

LSTM-Based Forecasting and Analysis of EV Charging Demand in a Dense Urban Campus

A Lightweight DL Model for Smart Grid Power Forecasting with Feature and Resolution Mismatch

Adaptive Online Learning with LSTM Networks for Energy Price Prediction

Abstain Mask Retain Core: Time Series Prediction by Adaptive Masking Loss with Representation Consistency

xTime: Extreme Event Prediction with Hierarchical Knowledge Distillation and Expert Fusion

Built with on top of