Energy Consumption Prognosticate In Built Environment Using GAN Approach

Abstract

Building accurate and robust models for forecasting energy consumption is a key objective in managing and operating smart buildings. While previous research has explored various load prediction techniques, the combined impact of data augmentation and machine learning on energy forecasting remains underexplored. To address this, the present study introduces a novel ensemble-based approach enhanced with Generative Adversarial Networks (GANs) for predicting energy usage in large commercial buildings.The proposed framework integrates multiple base models using a stacking ensemble method, while GANs are employed to learn the underlying data distribution and generate high-quality synthetic samples to augment the training dataset. This enriched dataset enables the model to train on a broader spectrum of scenarios, thereby improving robustness and generalization.The experimental evaluation investigates the effectiveness of the approach by employing three different GAN variants and measuring performance with metrics such as Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and the Coefficient of Variation of RMSE (CV-RMSE). Results highlight the practical potential of the proposed method, demonstrating its capability to deliver accurate energy consumption forecasts for real-world applications in smart building management.