A FUZZY EXPERT SYSTEM AND GENETIC ALGORITHMS TO SCORE THE SAFETY MANAGEMENT PERFORMANCE IN CONSTRUCTION SITES OF IRAN: CONSIDERING SAFETY CLIMATE

Construction is a dangerous and risky industry with a high accident rate. Safety, therefore, is deemed to be an important issue, and various agencies endeavor to control accidents through dierent approaches. Contractors procure worker compensation insurance to transfer these risks to insurance companies, but, as the safety management performance decreases, the risk of insuring this site for the insurance companies increases. Therefore, giving a score to the safety management performance in a construction site will help insurance companies to quote optimal premiums. If a contractor's real-time safety management system is robust, the contractor will get a cost-eective insurance cover, which will enable the contractor to compete better in tenders. Those contractors who do not have robust safety management systems in place will be penalized by higher premiums. Meanwhile, this approach will automatically set risky contractors aside and motivate contractors to invest in safety measures in their organizations. In this research, by considering the safety climate, personal experience factors, studying the literature and consulting with experts, the main factors eecting safety in the construction sites of Iran were recognized. Since all of these factors are linguistics, a fuzzy expert system was proposed for scoring their safety management performance. The parameters of the designed expert system, including Yager's t-norm, Yager's s-norm, Yager's complement, Yager's inference engine and Yager's defuzzication parameters, were optimized by genetic algorithms. The performance of the proposed fuzzy expert system for scoring the safety management performance in the construction sites of Iran was tested using the mean squared error (MSE) method. This method was utilized to quantify the dierence between values implied by the designed fuzzy expert system and the true values of the quantity being estimated. Testing results validate the eectiveness of the designed fuzzy expert system for scoring the safety management performance in construction sites of Iran.