CNN and GRU based Deep Neural Network for Electricity Theft Detection to Secure Smart Grid

In this paper, a Hybrid Deep Neural Network (HDNN) is proposed in this work, which is the combination of Convolutional Neural Network (CNN), Gated Recurrent Unit (GRU) and Particle Swarm Optimization (PSO), termed as CNN-GRU-PSO HDNN. In this paper, real time electricity consumption data of consumers is used, which is taken from an easily available online source, named as State Grid Corporation of China (SGCC). The original dataset consists of actual values along with the erroneous and missing values. The pre-processing steps are performed initially to refine the data. After that, feature selection and extraction are performed using CNN, which reduce both the dimensionality and the redundancy present in the dataset. Furthermore, the classification of provided data into honest and fake consumers is done using GRUPSO technique. The proposed HDNN model's performance is then compared with various benchmark techniques like Logistic Regression (LR), Support Vector Machine (SVM), Long Short Term Memory (LSTM) and GRU. The efficiency of the proposed model is validated using various performance parameters like Area Under the Curve (AUC), precision, accuracy, recall and F1-Score. The simulation results show that the proposed model outperforms the existing techniques in terms of ETD and class imbalanced issues. Moreover, the proposed model is also more robust and accurate than the existing methods.