Insulation Temperature Monitoring of Distribution Cables Based on Gaussian Process Classifier with Stationary Wavelet Transform

In today's era, with global energy demands escalating rapidly, reliable and efficient power distribution systems have become essential. Several methods have been proposed to monitor the temperature of cables which have scalability, limited-spatial resolution, and compliance issues. To overcome these drawbacks, a Gaussian Process Classifier (GPC) is employed to monitor the insulation temperature of distribution cables. To assesses this research, a Temperature Reading: IOT Devices (TR: IID) dataset is incorporated as an input for data preprocessing. Discrete Wavelet Transform (DWT) along with Hilbert Transform (HT), and Variational Mode Decomposition (VMD) are considered for preprocessing to filter background noise, to obtain frequencies and amplitudes of signals, and to reduce frequency overlapping. For feature extraction, a Stationary Wavelet Transform (SWT) is adopted to preserve temporal data, to reduce dimensionality, enhance accuracy, and for efficient computation. The GPC is leveraged as classifier for classification of insulation temperature monitoring to handle missing data, multi-dimensionality, and to enhance scalability, and interpretability. The proposed GPC model obtained high results in the context of precision 99.81%, recall 97.13 %, and F1 score 99.87 % and accuracy 99.72%, when compared to existing methods like Decision Tree (DT).