Safety risk assessment for connected and automated vehicles: Integrating FTA and CM-improved AHP

To reduce safety accidents from functional failures in connected and automated vehicles (CAVs), risk assessment and prevention are essential. However, traditional hazard analysis and risk assessment (HARA) methods suffer from limitations: insufficient quantitative assessment and inadequate consideration of ambiguity and uncertainty. To this end, we propose a quantitative risk assessment method for CAVs based on fault tree analysis (FTA) and cloud model (CM)-improved analytic hierarchy process (AHP). First, we use the golden section method and CM to refine the automotive safety integrity level (ASIL), representing ambiguity in level boundaries. Next, we incorporate potential functional failure paths and construct basic events for the FTA based on the failure modes of vehicle components. Meanwhile, the CM-improved AHP is applied to assess risks for each basic event, reducing uncertainty from subjective data. Finally, we combine the technique for order preference by similarity to an ideal solution (TOPSIS) with the conversion function to provide quantitative probabilities for the top event and perform a sensitivity analysis of basic events. A case study on a real open-source test vehicle shows that the proposed method quantifies the probability of automatic emergency braking (AEB) system failure and ranks the risk for each basic event. Compared to existing methods, it has significant advantages in the comprehensiveness and objectivity of risk assessment, providing more accurate information for risk prevention.