Authors :

Mariam Hassan and Youssef El-Shennawy

Address :

Faculty of Engineering, Alexandria University, 22 El-Gaish Road, Shatby, Alexandria, Egypt.

Faculty of Computers and Information, Helwan University, Ain Helwan, Cairo, Egypt.

Abstract :

The increasing deployment of smart grids has led to the generation of large-scale energy data, creating new opportunities for intelligent pattern mining to enhance energy efficiency and grid reliability. This study presents a hybrid optimization framework for mining meaningful patterns in smart grid energy data using Enhanced Binary Firefly Optimization (EBFO). Existing pattern mining techniques often struggle with high-dimensional data, noise, and low precision in pattern discovery, which limits their effectiveness in innovative grid environments. Additionally, traditional algorithms lack robustness when dealing with temporal variations and redundant features in energy consumption data. To address these challenges, this paper proposes a novel framework, the Binary Enhanced Firefly-based Pattern Miner (BE-FPM), integrated with a Thermal Image Denoising Autoencoder (TIDA). BE-FPM leverages an improved binary firefly algorithm with adaptive light intensity and movement strategies to explore the solution space for frequent pattern detection efficiently. Meanwhile, TIDA preprocesses smart meter readings by converting consumption patterns into thermal-like images and applying denoising autoencoding to reduce data noise and highlight meaningful structures. The proposed method is used on residential smart grid datasets for practical demand response analysis and load forecasting. By identifying accurate and noise-free consumption patterns, utilities can more effectively schedule energy distribution, reduce peak loads, and improve energy efficiency. Experimental results demonstrate that BE-FPM outperforms traditional mining approaches in terms of pattern accuracy, convergence speed, and noise resilience. This hybrid technique provides a promising direction for intelligent energy data analysis in future innovative grid applications.

Keywords :

Smart Grid, Pattern Mining, Binary Firefly Optimization, Thermal Image Denoising, Autoencoder, Energy Consumption Analysis, Demand Response.

DOI :