Study on prediction model of building construction safety accidents based on GA-SVM

In the Genetic algorithm and Support vector machine prediction method, the two methods are combined and applied to the construction safety accident prediction. Select main factors leading to construction safety accident, and then prediction model of construction safety accident is constructed. The model uses simulation cases as the training sample to determine the parameters in GA-SVM, and to carry on the forecast. The GA-SVM, SVM standard, BP neural network, fuzzy clustering and various methods are compared. The results show that: Prediction model based on GA - SVM has the rationality and effectiveness, and it highlights the characteristics of prediction results with high precision and strong stability, so there is practical value to construction.