The main factor that affects Loran-C navigation positioning accuracy is the ASF propagation delay generated during signal propagation. The accuracy of Loran-C positioning can be improved by correcting the ASF. First, the Loran-C signal coverage area was rasterized and then the relationship between earth conductivity, atmospheric refractive index, temperature, humidity and ASF correction was found through training of neural network. Finally, the theoretical predicted value is compared with the actual value, and the results show that method is effective.