Comprehensive Multi-Sheet Simulation and Diagnostic Dataset for AI-Driven Safety Analysis in Complex Engineering Systems

This dataset provides a comprehensive, multi-layered evidence base supporting the development, validation, and evaluation of artificial intelligence driven simulation models for safety analysis, diagnostics, and decision support in complex engineering systems. The dataset was designed to ensure methodological rigor, traceability, and reproducibility, and to comply with international expectations for data transparency in high-impact engineering research. The database is organized as a structured, multi-sheet Excel repository integrating raw signals, engineered features, simulation outputs, validation results, and compliance diagnostics. It includes more than 1,500 validated data instances, expanded through window-based segmentation and simulation scenarios to exceed tens of thousands of analytical records, enabling robust statistical analysis and machine learning training under realistic operational variability. The dataset incorporates multiple interconnected components. Subject-level and system-level sheets define experimental configurations, acquisition parameters, preprocessing pipelines, and target performance thresholds. High-resolution windowed data capture both nominal and anomalous operating conditions, supporting supervised and unsupervised learning tasks such as fault detection, anomaly identification, and risk classification. Diagnostic labels reflect real-world imbalance scenarios common in safety-critical systems. Simulation sheets implement controlled perturbations including noise injection, signal distortion, operational stress, and boundary-condition variation. These simulators allow systematic evaluation of model robustness, sensitivity, and generalization under adverse or extreme conditions. Validation sheets report cross-scenario and leave-one-out evaluation metrics, including accuracy proxies, sensitivity, specificity, and consistency indicators, enabling transparent performance assessment. Additional sheets document energy, efficiency, and operational constraints relevant to cyber-physical and embedded systems, as well as cybersecurity and compliance mappings aligned with engineering governance, safety management, and regulatory auditing requirements. Automated diagnostic flags and pass or fail indicators are embedded throughout the dataset to support audit-ready analysis. This dataset is intended for use by researchers, engineers, and practitioners working in artificial intelligence, safety engineering, cyber-physical systems, digital twins, and industrial process monitoring. It supports reproducible experimentation, benchmarking of AI-based diagnostic models, and methodological extension to other safety-critical engineering domains.