Case-based reasoning for automated safety risk analysis on subway operation: Case representation and retrieval

Abstract

Risk analysis is a core process of safety management. By analysis of the precursors appeared on the subway system, the safety risk could be identified and effective measures could be taken to avoid accident. This paper proposes the use of Case-Based Reasoning (CBR), which combines case representation and retrieval to analyze safety risk. The innovation of the proposed method lies in the structure of the semantic networks which contain sub-concepts to describe all the possible precursors from workers, physical system and environment. The flexibility of the method allows multiple forms of qualitative precursors to be used when only qualitative aspects are known. Details have been provided on the case representation scheme and automated retrieval mechanism. A real-world example demonstrates the feasibility of the proposed method. The results of the case study show that the similarity between the stored case (HET1) and the input case is 0.50, while the similarity between the stored case (HET3) and the input case is only 0.22. Given a similarity baseline, the safety risk and safety measures of the input case could be automated analyzed according to the retrieved case. The proposed method increases the applicability of conventional CBR analysis to many real-world settings, where the safety risk analysis depends on more qualitative precursors.