Faulty Behaviors Simulation in Industrial Cyber-Physical Systems for Safety Analysis

Industrial Cyber-Physical Systems (ICPS) represent a convergence of digital, physical, and networked domains, where ensuring functional safety is of paramount importance. This thesis presents a comprehensive methodology for simulating and analyzing faulty behaviors in ICPS, with a particular focus on multi-domain fault modeling, injection, and detection. By leveraging physical analogies, especially among electrical, mechanical, and thermal domains, the work introduces an innovative approach to extend standardized electrical fault injection techniques (e.g., ISO 26262, IEEE 2427-2025) to non-electrical domains. The proposed methodology enables the derivation of equivalent fault models across domains by exploiting analogies such as impedance and mobility, facilitating the simulation of complex fault scenarios in heterogeneous systems. The approach is validated through multiple case studies, including DC motors, MEMS accelerometers, and lithium-ion battery packs (e.g., the Tesla Model S), which are modeled using Verilog-AMS and SystemC AMS. These models incorporate electrical, mechanical, and thermal behaviors, allowing for accurate fault injection and behavioral analysis. Furthermore, the thesis explores fault detection strategies based on contract-based monitoring and Time-Sensitive Behavioral Contracts (TSBCs), extending the analysis to software faults and control systems. The integration of Unreal Engine for immersive simulation and visualization, along with the development of a human digital twin framework, demonstrates the applicability of the methodology in Industry 4.0 contexts. The results highlight the effectiveness of the proposed multi-domain fault modeling and simulation framework in enhancing the robustness, safety, and diagnosability of ICPS. This work lays the foundation for future research in fault isolation, predictive maintenance, and the integration of real-time monitoring systems in complex industrial environments.