Abstract: Failures in safety-critical software will lead to catastrophic consequences. Software failure modes and effect analysis(SFMEA) is a safety analysis method which is commonly used in industry. However, the SFMEA method relies on manual analysis, thus it has poor reusability and low accuracy. To solve these problems, this paper proposes a creative SFMEA method combined with ontology and rule reasoning. Firstly, the SFMEA ontology is developed to accurately describe the software failure knowledge. Secondly, rules are built to describe the failure analysis processes using semantic web rule language(SWRL). According to the failure causes, the reasoning engine can analyze the failure effects automatically. At last, a case study of engine flameout-state module analysis confirms the feasibility and practicality of the research.

Key words: safety-critical software, software safety, SFMEA, ontology, rule reasoning