Tube Loss based Deep Networks For Improving the Probabilistic Forecasting of Wind Speed

TL;DR

This paper introduces a novel deep learning approach using Tube loss for probabilistic wind speed forecasting, achieving narrower and more reliable prediction intervals without distribution assumptions.

Contribution

It proposes a model-agnostic Tube loss function integrated with deep architectures like LSTM, GRU, and TCN for improved wind speed uncertainty quantification.

Findings

Models produce narrower prediction intervals.

Enhanced reliability of probabilistic forecasts.

Validated on datasets from three different locations.

Abstract

Uncertainty Quantification (UQ) in wind speed forecasting is a critical challenge in wind power production due to the inherently volatile nature of wind. By quantifying the associated risks and returns, UQ supports more effective decision-making for grid operations and participation in the electricity market. In this paper, we design a sequence of deep learning based probabilistic forecasting methods by using the Tube loss function for wind speed forecasting. The Tube loss function is a simple and model agnostic Prediction Interval (PI) estimation approach and can obtain the narrow PI with asymptotical coverage guarantees without any distribution assumption. Our deep probabilistic forecasting models effectively incorporate popular architectures such as LSTM, GRU, and TCN within the Tube loss framework. We further design a simple yet effective heuristic for tuning the $\delta$ parameter of the Tube loss function so that our deep forecasting models obtain the narrower PI without compromising its calibration ability. We have considered three wind datasets, containing the hourly recording of the wind speed, collected from three distinct location namely Jaisalmer, Los Angeles and San Fransico. Our numerical results demonstrate that the proposed deep forecasting models produce more reliable and narrower PIs compared to recently developed probabilistic wind forecasting methods.