A novel deep learning technique to detect electricity theft in smart grids using AlexNet

Abstract

Electricity theft (ET), which endangers public safety, interferes with the regular operation of grid infrastructure, and increases revenue losses, is a significant issue for power companies. To find ET, numerous machine learning, deep learning, and mathematically based algorithms have been published in the literature. However, these models do not yield the greatest results due to issues like the dimensionality curse, class imbalance, inappropriate hyper-parameter tuning of machine learning, deep learning models etc. A hybrid DL model is presented for effectively detecting electricity thieves in smart grids while considering the abovementioned concerns. Pre-processing techniques are first employed to clean up the data from the smart meters, and then the feature extraction technique, AlexNet is used to address the curse of dimensionality. An actual dataset of Chinese smart meters is used in simulations to assess the efficacy of the suggested approach. To conduct a comparative analysis, various benchmark models are implemented as well. This proposed model achieves accuracy, precision, recall, and F1-score, up to 86%, 89%, 86%, and 84%, respectively.