Power Grid Communication Optical Cable Monitoring System Using Bidirectional Gated Recurrent Unit and Adaptive Moment Estimation with Weight Decay

In this era, the power grid communication optical cable monitoring system is crucial for ensuring the reliable function and security of contemporary power grids, where optical cables are gradually utilized for high-speed communication among numerous components. Previous researchers suggested various tradition methods but still failed to monitor health and integrity of optical cables which is a significant challenge leading to probable communication failures. Therefore, this research suggested a hybrid approach using various techniques such as Bidirectional Gated Recurrent Unit (Bi-GRU) and Adaptive Moment Estimation with Weight Decay (AdamW) optimization model which is validated through China Southern Power Grid dataset. Then, hybrid noise reduction model namely Median Filter (MF) and Savitzky Golay Filter (SGF) are incorporated to resolve the issue of imbalance data. Further, statistical feature extraction techniques are employed such as mean, variance, skewness and kurtosis by which optimal features are selected on the basis of resulted values. Finally, a hybrid model using Bi-GRU-AdamW optimization model is incorporated for monitoring the power grid communication cable. The proposed Bi-GRU-AdamW model has acquired greater results in the terms of precision of 98.3%, recall of 93.7%, mean Average Precision (mAP) of 92.8% when compared to You Only Look Once version 5 (YOLOv5) model.