2025, Vol. 6, Issue 2, Part A
Machine learning-based load forecasting in data-driven power communication systems
Author(s): Olivia Johnson, Ethan Williams, Mia Brown and Noah Davis
Abstract: The increasing integration of digital communication technologies in modern power systems has transformed traditional grids into highly data-driven infrastructures, demanding accurate and adaptive load forecasting mechanisms. This study presents a machine learning-based load forecasting framework that explicitly incorporates communication-layer characteristics such as latency, jitter, and packet loss into the prediction process. Using a hybrid model combining Long Short-Term Memory (LSTM) networks and Support Vector Regression (SVR), the research evaluates forecasting performance under varying communication conditions within a simulated smart grid environment. Comparative analysis with conventional models including ARIMA, Feed-Forward Neural Network (FFNN), and XGBoost demonstrates that the hybrid LSTM-SVR model achieves the highest forecasting accuracy and resilience, with the lowest Mean Absolute Percentage Error (MAPE) of 1.9% under ideal conditions and 3.1% under severe network disturbances. Statistical testing using bootstrap and permutation methods confirmed the hybrid model’s superiority with a significant p-value (< 0.01). Furthermore, the study highlights how communication impairments can amplify prediction errors if not addressed during model design. The results validate the hypothesis that coupling load dynamics with communication behavior enhances forecasting reliability and operational efficiency. By integrating deep learning with communication-aware modeling, this framework provides a scalable approach for real-time energy management, demand response optimization, and digital twin applications in smart grids. The proposed system offers a significant step toward achieving predictive stability, cost efficiency, and data resilience in next-generation power communication networks.
Pages: 51-56 | Views: 168 | Downloads: 63
Download Full Article: Click Here
How to cite this article:
Olivia Johnson, Ethan Williams, Mia Brown, Noah Davis. Machine learning-based load forecasting in data-driven power communication systems. Int J Electr Data Commun 2025;6(2):51-56.
