Risk Early-Warning Framework for Government-Invested Construction Project Based on Fuzzy Theory, Improved BPNN, and K-Means

Government-invested construction project (GICP) has a great significance to social and economic development but suffered many risks due its large scale, huge investment, long period. The in GICP are complex so as lead the failure; it is extremely urgent take risk management of GICP. This study establishes early-warning framework help managers understand threat advance, which supports them make proper strategies for control. whole can be concluded three parts: information collection, data processing, result prediction. Firstly, 16 factors identified. To express hesitance human decision reduce loss quantification, hesitant fuzzy linguistic term set (HFLTS) triangular number (TFN) used collect experts’ transform into numerical value. And then, these inputs simplified five based on principal component analysis (PCA), decreasing impact redundancy early-warning. Meanwhile, warning level divided K-means, avoids subjectivity experience decision. Further, backpropagation neural network optimized by genetic algorithm (GA-BP) complete simulation 75 groups questionnaire train 10 test set. validation proposed been verified with an average relative error 7.2% absolute 3.91. Finally, corresponding suggestions prevent control different put forward.