FDIA-SyntheticMicroGrid: Complex Spatio-Temporal False Data Injection Attack (FDIA) Dataset for Microgrid Systems

The dataset contains two million rows of data generated for the IEEE-5, IEEE-9, IEEE-14, and IEEE-30 bus architectures, comprising both attack and benign samples, with each bus system contributing 0.5 million records. The attack classes consist of thirty different combinations of False Data Injection Attacks (FDIAs), organized into three major categories based on injection methods (i.e., Additive, Deductive, Combined), attack strategies (i.e., random_node or vulnerable_node), and behavioral patterns (i.e., step, ramp, pulse, random, or replay). The dataset was generated using the MATLAB-based MATPOWER power flow simulation environment. Benign samples were produced through power flow simulations under nominal microgrid operating conditions, whereas attack samples were generated under various microgrid operating scenarios and attack conditions. Detailed descriptions of each label are provided in the legend_label.txt file. This dataset was developed to facilitate the evaluation of Artificial Intelligence-, Deep Learning-, and Quantum Machine Learning-based FDIA detection models in microgrid environments.