Construction and Empirical Research of Engineering Data Prediction Model Based on BP Neural Network

At present, it is very important to select the data prediction algorithm and model which can meet the demand of engineering construction project for the whole construction stage. By analyzing and studying the basic theory of BP neural network model, this paper finds that the model has the ability to deal with nonlinear network relations of multi-input and multi output factors in parallel, and trains and learns network relations through the back propagation algorithm of output layer, hidden layer and input layer data. At the same time, it is found that the processing ability of multi-input and multi output nonlinear network relationship is exactly consistent with the multi-factor nonlinear mapping relationship of engineering data. Finally, taking bridge engineering data as an example, BP neural network toolbox in matlab software is used to train and learn the collected sample data with excitation function, training function, learning function and performance function, and then the simulation data of network model is compared with the actual sample data, as well as the error analysis of the new sample data. The prediction function of the generated BP neural network model on engineering data is proved, and then the prediction method of engineering data can be used for reference.