By using the artificial neural network method based on the Lick line index to develop atmospheric physical parameters measurement, Kurucz synthetic spectra preprocessing was used to adapt to the requirements of the final LAMOST spectral data, Lick line index with the corresponding atmospheric physical parameter was input, and training artificial neural network was used to get the training model through DR8 spectroscopic data for testing. The artificial neural network is aimed to achieve the best experimental effect by adjusting the relevant parameters. The experimental results show that through the artificial neural network method to get the atmospheric physical parameters is feasible.