Load spectrum measurement is a necessary step in aircraft development, and load inversion model directly determines the accuracy of load spectrum measurement. Based on the finite element method, the strain response database of a UAV wing structure is obtained, and two typical load inversion models, multiple linear regression and popular neural network under the background of big data, are established. The prediction accuracy of the two load inversion models with different sample size and sample quality is analyzed, the advantages and disadvantages of the two load inversion models are compared, and the trial range of the linear regression equation and the neural network load inversion model is obtained. The analysis method and analysis results can provide a basis for the selection of the load inversion model in the model.