Research on Quality of Prefabricated Construction Components Based on MIV-BP Neural Network Optimization Algorithm

This study proposes a method of construction component quality prediction on the basis of the MIV-BP neural network optimization. It uses MIV to display the change of the proportion of the input value in the neural network. By selecting the input value of the BP neural network model, it selects the variables that have a great impact on the model. In this way, the number of input parameters required for modeling can be reduced and the training accuracy of the model can be improved. Among them, MIV can carry out numerical transformation on an output parameter, and then predict through the BP neural network shape of the component after the change. By comparing the two prediction results before and after the change, the order of importance of each input parameter on the correlation of output parameters can be obtained, and the characteristic parameters that have less impact on the output results can be removed from the model, so as to screen the input parameters. It improves the complexity of BP neural network matrix when the number of input values increases significantly, and greatly builds up the calculation potency of the model. It helps to increase the reliable of the prediction of the production quality of PCC.