Dynamic Line Rating using Hyper-local Weather Predictions: A Machine Learning Approach

TL;DR

This paper presents a machine learning-based method for dynamic line rating that uses hyper-local weather forecasts and topographical data to improve accuracy and reliability in renewable energy transmission.

Contribution

It introduces a novel ML approach combining hyper-local weather predictions and topographical data for DLR estimation, overcoming sensor installation challenges.

Findings

Effective DLR prediction in a real-world Estonia case study

Enhanced accuracy with topographical data integration

Confidence intervals improve risk management in DLR assessments

Abstract

Dynamic Line Rating (DLR) systems are crucial for renewable energy integration in transmission networks. However, traditional methods relying on sensor data face challenges due to the impracticality of installing sensors on every pole or span. Additionally, sensor-based approaches may struggle predicting DLR in rapidly changing weather conditions. This paper proposes a novel approach, leveraging machine learning (ML) techniques alongside hyper-local weather forecast data. Unlike conventional methods, which solely rely on sensor data, this approach utilizes ML models trained to predict hyper-local weather parameters on a full network scale. Integrating topographical data enhances prediction accuracy by accounting for landscape features and obstacles around overhead lines. The paper introduces confidence intervals for DLR assessments to mitigate risks associated with uncertainties. A case study from Estonia demonstrates the practical implementation of the proposed methodology, highlighting its effectiveness in real-world scenarios. By addressing limitations of sensor-based approaches, this research contributes to the discourse of renewable energy integration in transmission systems, advancing efficiency and reliability in the power grid.