Boost Converter Dataset for Cyber-Attack Detection and Analysis

Dataset Description Overview The Boost Converter Cyber-Attack Benchmark Dataset is a labeled dataset developed to facilitate research on cyber-attack detection, classification, and resilience assessment in DC–DC boost converter systems. The dataset was generated using a detailed MATLAB/Simulink model of a boost converter operating under both normal and malicious conditions. It captures the dynamic electrical behavior of the converter during different cyber-attack scenarios and provides a standardized benchmark for evaluating data-driven detection and classification methods. The dataset contains 9,996 labeled samples, each represented by eight numerical features extracted from the converter's electrical responses. Every sample is assigned to one of five operating conditions, including one normal state and four distinct cyber-attack categories. Dataset Composition Total samples: 9,996 Number of features: 8 Number of classes: 5 Data type: Numerical Format: CSV Simulation platform: MATLAB/Simulink Class Labels Label Operating Condition 0 Normal Operation 1 False Data Injection (FDI) Attack 2 Replay Attack 3 Denial-of-Service (DoS) Attack 4 Covert Attack The dataset contains one majority class representing normal converter operation and four attack classes corresponding to different cyber-attack mechanisms. Feature Description Each record consists of eight numerical features extracted from the boost converter measurements. These features characterize the electrical behavior of the converter during normal and abnormal operating conditions. The dataset includes: Output voltage characteristics Inductor current characteristics Dynamic voltage variation Statistical electrical descriptors Time-domain operating characteristics These features capture both steady-state and transient behaviors of the converter, allowing researchers to distinguish different attack scenarios. Data Generation Methodology The dataset was generated using a high-fidelity MATLAB/Simulink model of a DC–DC boost converter. The converter was operated under normal conditions as well as multiple cyber-attack scenarios. During simulation, electrical measurements were continuously recorded and processed to extract representative features from each operating interval. For every operating condition, the extracted feature vectors were assigned the corresponding class label, resulting in a structured supervised learning dataset suitable for classification, anomaly detection, feature analysis, and benchmarking studies. Potential Applications The dataset can be used for a wide range of research topics, including: Cyber-attack detection in power electronic converters Intelligent monitoring of DC–DC converters Smart grid cybersecurity Renewable energy system protection Feature selection and explainable AI studies Benchmark evaluation of classification algorithms Deep learning and traditional machine learning comparisons Fault diagnosis and anomaly detection Benchmark Value This dataset serves as an open benchmark for evaluating cyber-attack detection techniques in boost converter systems. It enables fair comparison among different algorithms by providing a common labeled dataset with multiple attack categories and standardized operating conditions. Researchers may use this dataset to compare classification performance, robustness, computational efficiency, feature importance, and generalization capability across various analytical methods. Citation If you use this dataset in your research, please cite the associated publication and the Zenodo dataset DOI.