The propagation of electromagnetic waves in evaporation duct follows Maxwell's law. Theoretically, the electromagnetic wave propagation process can be solved in case of knowing electromagnetic wave source characteristics, propagation medium characteristics and boundary conditions. However, the calculation of evaporation duct profiles and Maxwell's equations is complex and slow. In most cases, it can only be solved by using numerical methods. A simple, fast and approximately accurate scheme of path loss calculation is needed to overcome this difficulty. Therefore, deep neural network is used to estimate the path loss of electromagnetic wave propagation in evaporation duct in this paper.