An Automated Accident Causal Scenario Identification Method for Fully Automatic Operation System Based on STPA.

Fei Yan,Junqiao Ma,Mo Li,Ru Niu,Tao Tang

DOI: https://doi.org/10.1109/access.2021.3050472

IF: 3.9

2021-01-01

IEEE Access

Abstract:Accident causal scenario can describe the process logic of the accident clearly and concretely from the perspective of the control mechanism. Only by improving the quality of the causal scenario can the effective control measures be taken. Combining the technical characteristics of the fully automatic operation (FAO) system, the paper proposes an automated accident causal scenario identification method for FAO system based on the System-Theoretic Process Analysis (STPA) method. Aiming at the problem that there are too many layers in the hierarchical control structure diagram of STPA method, which makes it impossible to effectively trace the cause and the problem that the basic control structure model only contains the control structural information and lacks the cause information, a new basic control structure model is defined to model multiple control processes in time sequence, and then the paper extends it from four aspects: control action, input variables, external disturbance, and synchronous timing to add more system cause information. For the lack of a unified standard description problem for the causal scenario, a four-stage causal scenario description method is defined, this paper has developed the first timing, non-first timing, synchronous timing, and external disturbance causal scenario search rules to ensure the automatic identification of the causal scenarios. Applying the automated safety analysis method to the case study of the operational scenarios of parking in a station of Beijing Yanfang Line, the automated identification of related causal scenarios is successfully completed through the Auto-STPA platform, and corresponding safety requirements are added. The feasibility of the method and the applicability to the analysis of operational scenarios are verified.