Predictive models for short-term load forecasting in the UK's electrical grid

Area of Science:

• Electrical Engineering
• Computer Science
• Energy Systems

Background:

• Global deployment of electric vehicles (EVs) is accelerating the energy transition.
• Increased EV adoption raises concerns about potential strain on electricity grids due to higher demand.
• Accurate short-term load forecasting is essential for efficient grid planning, operation, and control.

Purpose of the Study:

• To develop and evaluate robust machine learning models for short-term load forecasting in the UK's power system.
• To compare the performance of Support Vector Regression (SVR), Artificial Neural Networks (ANN), and Gaussian Process Regression (GPR) for load prediction.
• To assess the accuracy of forecasting models for both half-hourly and hourly electricity demand.

Main Methods:

• Utilized net imports data from the UK's power system spanning 2010-2020.
• Applied machine learning techniques including Support Vector Regression (SVR), Artificial Neural Networks (ANN), and Gaussian Process Regression (GPR).
• Evaluated model performance using metrics such as Root Mean Square Error (RMSE), Mean Absolute Prediction Error (MAPE), Mean Absolute Deviation (MAD), and Correlation of Determination (R2).

Main Results:

• Support Vector Regression (SVR) demonstrated the highest accuracy for half-hourly load forecasts, achieving an R-value of 99.85%.
• Artificial Neural Networks (ANN) provided the best performance for hourly load forecasts, with an R-value of 99.71%.
• Gaussian Process Regression (GPR) also showed strong predictive capabilities for both forecast horizons.

Conclusions:

• Machine learning methods, including SVR, ANN, and GPR, are highly reliable and precise for short-term load forecasting.
• The SVR model is particularly effective for half-hourly predictions, while the ANN model excels in hourly predictions.
• Accurate load forecasting is vital for integrating renewable energy and managing the evolving demands of e-mobility.