Abnormal Electricity Consumption Detection Method For Smart Grid Using Fusion Matrix Completion And Improved Clustering Algorithm

Abstract

This project presents an intelligent abnormal electricity consumption detection method for smart grid systems using Fusion Matrix Completion and Improved Clustering Algorithm techniques integrated with machine learning and deep learning models. The main objective of the proposed system is to accurately identify abnormal or anomalous electricity consumption patterns in smart grid environments to improve energy management, reduce power theft, and enhance grid security. The system utilizes a smart grid dataset containing parameters such as voltage, current, frequency, power flow, packet count, byte count, device ID, and attack type to analyze consumer energy behavior and detect suspicious activities. Initially, the dataset is preprocessed using label encoding, normalization, and missing value handling techniques to improve data quality and model efficiency. Fusion matrix completion techniques are applied to handle incomplete and inconsistent data, while the improved clustering mechanism groups similar consumption patterns for better anomaly identification. The processed data is then trained using multiple machine learning algorithms including Random Forest, Naive Bayes, Decision Tree, and a deep learning-based Long Short-Term Memory (LSTM) model to compare prediction performance.