Abstract: Applying artificial neural network technology and using MATLAB software to design BP network model to establish the complex non-linear relationship between technical parameters of polyurethane formulations and tensile strength can predict the corresponding tensile strength value of the specified technical parameters of polyurethane formulations. The research results demonstrate that neural network technology can be used in the design of polyurethane formulation and the established neural network model can more or less correctly reflect the regularity between technical parameters of polyurethane and its tensile strength. Model prediction error for tensile strength can be controlled within 4.60 which shows the model has a guiding significance in the experiment.