Research on Predictive Model of Prefabricated Building Safety Risk Based on RS-SVR

To improve the prediction precision of safety risk for prefabricated building construction, a prediction model based on Rough Set (RS) and Support Vector Regression (SVR) is studied. The mechanisms of RS and SVR are combined. Considering the distinguishing features of prefabricated building construction, a risk comprehensive assessment index system is constructed. It consists of five aspects: human, equipment and materials, management, surrounding and technology. The detail processes or steps of data acquisition and preprocessing, attribute reduction by RS, training by SVR are described. The safety risk of prefabricated building is predicted by the prediction model based on RS-SVR. The application result shows that compared with the traditional BP neural network model, above RS-SVR model owns both faster computational velocity and higher prediction precision. For prefabricated building construction, it provides a novel way to predict the safety risk.