Fuzzy probabilistic expert system for occupational hazard assessment in construction. (March 2017)

• Record Type: Journal Article
• Authors: Amiri, Mehran; Ardeshir, Abdollah; Fazel Zarandi, Mohammad Hossein
• Is Part Of: Safety science. Volume 93(2017)
• Page Start: 16
• Page End: 28
• Publication Date: 2017-03
• Subjects: Construction -- Fuzzy-probability logic -- Rule base expert system -- Safety risk assessment; Industrial accidents -- Periodicals; Accident Prevention -- Periodicals; Safety -- Periodicals; Travail -- Accidents -- Périodiques; 363.11
• DOI: 10.1016/j.ssci.2016.11.008
• Languages: English
• ISSNs: 0925-7535

Abstract

Highlights: Major improvement is achieved by merging randomness into ORA problem. Fuzzy probabilistic rules are extracted from association rules and decision trees. The methodology prevents from losing information when using data mining methods. More accuracy is achieved by considering root factors in designing the model. By an intensive validation process, the model was validated with a very good grade. Abstract: Considering the extensive growth of the construction industry in developing countries, the trend of occupational accidents in this sector is growing in recent years. In this regard, developing a hazard management process with a proactive vision makes it possible to identify and prioritize risky points in construction sites and apply preventive measures. Hence, in this paper, a fuzzy probabilistic rule-based expert system is developed for occupational hazard assessment. A fuzzy probabilistic system permits us to model uncertainties related to accident databases and the randomness due to environmental, natural, or time changes. Merging randomness into the occupational risk assessment problem in the construction industry enables the authorities to manage hazards proactively and brings about some practical benefits. The proposed fuzzy probabilistic model benefits from a rule base generated based on fuzzy risk-based statistical and data mining analyses of accident database along with a comprehensive literature review and interviews with experts. This model is tested on four major construction case studies. Through an intensive validation process, the model was successfully analyzed and ranked the risks of different types. The results are encouraging and the model can be implemented in different construction projects.