Detecting Non-Technical Losses in the Energy Sector using MLPGRU: An Anomaly Detection Approach

Electricity theft (ET) is a major problem for smart grids and costs the world's power companies a lot of money. The dimensionality curse and the uneven distribution of data on energy use make conventional electricity theft detection (ETD) algorithms challenging. This work offers a unique hybrid approach that combines the Multi-Layer Perceptron (MLP) and Gated Recurrent Unit (GRU) to overcome these issues. The State Grid Corporation of China's (SGCC) dataset serves as the input for the GRU network's analysis of smart meter data, while the MLP network uses auxiliary data to examine benign aspects of power consumption statistics. With scores of 87% accuracy, 91% AUC, 90% recall, 87% precision, and 89% F1-Score, the suggested approach shows promise. The suggested model performs better than current methods for detecting power theft.