This paper innovatively proposes a hybrid intelligent system combining rough set approach and artificial neural network (ANN) that predicts the safety performance of construction site for breaking through the limitations of conventional method. Redundant attribute is removed with no information loss through rough set approach, by which the reduced information table is obtained. And then, this reduced information is used to develop classification rules and train neural network to infer appropriate parameters. The rules developed by rough set analysis show the best prediction accuracy if an empirical does match any of the rules. The effectiveness of our methodology was verified with an empirical study that compared neural network approach with the hybrid approach. And the results show that this method can be an effective tool to predict the safety performance of construction project sites, which is useful to provide a scientific basis for the management and decisions of accident prevention.

This paper innovatively proposes a hybrid intelligent system combining fuzzy comprehensive assessment approach and artificial neural network (ANN) that predicts the safety performance of construction site for breaking through the limitations of conventional method. And also inducts the sensibility analysis to discriminate the importance of each index in the assessment index system. The effectiveness of our methodology was verified with an empirical study by using the data of 30 construction sites we investigated. The results indicates that the Fuzzy-ANN model can accurately realize the non-linear reflection from input index to output index, and it can avoid the random and subjectivity when determining the indices weight. And the results show that this method can be an effective tool to predict the safety performance of construction projects sites, which is useful to provide a scientific basis for the management and decisions of accident prevention.