Forecasting Energy Demand with Deep Temporal Convolutional Networks in Smart Grid Applications

Authors:
R. Raguraman, K. Sakthivel

Addresses:
Department of Electrical and Electronics Engineering, Bharath Institute of Higher Education and Research, Chennai, Tamil Nadu, India.

Abstract:

A grid is a power transmission system that interconnects power sources and distributes electricity to specific users through a high-voltage transmission approach and small-scale, low-voltage grid distribution. Energy demand forecasting involves predicting the electricity a particular user will consume based on electrical device usage and load demand, while avoiding impaired power connections due to overloading or under-processing of power systems. In previous work, challenges such as inaccurate or partial historical data, high error rates due to non-linear power supply, and increasing consumer demand on a day-by-day basis have been identified. To solve the problem, a Deep Temporal Convolutional Network (DTCN) is proposed for the Power Demand Forecasting approach in the Smart Grid application. DTCN is a deep learning architecture that captures temporal dependencies by preserving the input sequence and enabling parallel processing across time steps in each transmission. Forecasting energy demand helps grid operators identify periods of high consumption, optimizing power generation and distribution. The Simulation calculates RMSE parameters, including MAPE, which represents the percentage of absolute errors compared to the exact values. The Simulation achieves highly accurate forecasting using widely accessible deep learning procedures through open-source platforms.

Keywords: Forecasting Energy Demand; Load Demand; High-Voltage Transmission; Power Transmission System; Smart Grid Application; Open-Source Platforms; Power Demand Forecasting.

Received on: 12/11/2024, Revised on: 21/01/2025, Accepted on: 18/03/2025, Published on: 05/10/2026

DOI: 10.69888/FTSES.2026.000797

FMDB Transactions on Sustainable Energy Sequence, 2026 Vol. 4 No. 2, Pages: 110-122

  • Views : 55
  • Downloads : 8
Download PDF