Weighted Recursive Graph Color Coding for Enhanced Load Identification

In the pursuit of high-precision load identification, traditional methodologies grapple with significant drawbacks, including low recognition rates, intricate signature construction, and narrow applicability. This study introduces a novel approach employing weighted recursive graph (WRG) color coding to surmount these challenges. Power consumption data, procured from advanced load monitoring devices, undergo extraction of single-cycle currents, which are then subjected to dimensional reduction via Piece-wise Aggregate Approximation (PAA). In a transformative step, these currents are encoded into load signatures through the recursive graph time series methodology, culminating in the generation of WRG images. An AlexNet neural network model is engaged to distil and assimilate the distinctive features of the WRG images. The simulation results indicate that the identification rate can exceed 97%. Additionally, an experimental platform was set up to verify the method proposed in this paper, and the results show that the actual identification rate can reach over 96%. Both the simulation results and experiments fully demonstrate that the proposed identification method has a high accuracy. This method not only sets a new standard in non-intrusive load identification but also enhances the generalization of load signature applicability across diverse scenarios.

@article{Zhang2024,
  title     = {Weighted Recursive Graph Color Coding for Enhanced Load Identification},
  journal   = {International Journal of Advanced Computer Science and Applications},
  volume    = {15},
  number    = {5},
  year      = {2024},
  publisher = {The Science and Information Organization},
  author    = {Li Zhang and Hengtao Ai and Yuhang Liu and Shiqing Li and Tao Zhang},
  doi       = {10.14569/IJACSA.2024.01505113},
  url       = {https://doi.org/10.14569/IJACSA.2024.01505113}
}