The integration of smart grid technology offers numerous benefits, including enhanced reliability, efficiency, and the capacity to integrate renewable energy sources. However, it also introduces vulnerabilities to cyber-attacks, such as False Data Injection Attacks (FDIAs). This study aims to investigate the detection of FDIAs using machine learning models, namely Random Forest, CatBoost, LightGBM, and XGBoost. We evaluate the performance of these models based on metrics such as accuracy, precision, recall, and training time. Our findings indicate that LightGBM and XGBoost demonstrate superior performance with high accuracy and low false positive rates. Moreover, we address the challenges associated with data imbalance and propose future directions for improving FDIA detection in smart grids.