A fuzzy system hazard analysis approach for human-in-the-loop systems

Abstract Objective The objective of this study was to show how fuzzy representation of hazard risk (membership functions and risk bands) could be used along with quantification of human-automation reliability to produce precise risk assessments and broad sets of control measures. Background Although some prior research has integrated fuzzy logic into several hazard analysis techniques and identified advantages in terms of risk estimation and recommended control measures, in applying systems hazard analysis (SHA), analysts still rely on discrete risk assessment approaches. The SHA approach is the only hazard analysis technique that accounts for hazards due to interactions among system components. However, the original method does not consider the impact of human-automation reliability in risk exposure calculations. Method The Mamdani max-min inference method was used to calculate a “risk-reliability” (R-R) score based on a fuzzy definition of frequency of hazard occurrence, severity of hazard outcomes, and system reliability. The application of the proposed model is presented in the context of a complex-human-in-the-loop system using the MATLAB® fuzzy logic toolbox. Results The fuzzy R-R score provided a precise risk assessment and led to broader set of control measures, as compared to use of the original SHA in a construction safety application. The proposed fuzzy inference system can be used by analysts through a graphical user interface. Conclusion The enhanced SHA supports broader safety control recommendations and provides more comprehensive results than the original technique. However, the new approach needs to be further evaluated for reliability in different applications.