Focusing on the engineering challenge of gas turbine engine blades in aerospace engineering subjected to complex and extreme multi-field coupled loads during service, we establish a thermo-mechanical decoupling method for standard tensile specimens in this research. Based on non-contact strain and temperature measurement techniques, the finite element method is employed to reconstruct the specimen surface temperature field and analyze thermal strain fields for decoupling. Furthermore, the proposed method is empowered with neural network model. Using measured temperature and Digital Image Correlation (DIC) data as inputs, and employing the Mean Squared Error to evaluate model performance, a thermo-mechanical decoupling neural network analysis model is developed. Simulation tests demonstrate that the neural network model enables rapid and accurate thermo-mechanical decoupling calculations.