Abstract: Path loss represents the attenuation of wireless signal power during propagation, caused by distance, obstacles, and other factors. It is a key parameter in wireless communication system design and performance evaluation. However, in complex electromagnetic environments such as substations, densely deployed power equipment leads to significant multipath effects, making traditional channel modeling methods inadequate for accurate signal propagation characterization. In order to improve the prediction accuracy of path loss in complex environments such as substations, an environment-resolved fusion prediction method for substation scenarios is proposed. First, based on the shape characteristics of scatterers in the environment, buildings and various types of power equipment in the substation scene are classified into regular scatterers and non-regular scatterers. Second, a three-dimensional simulation environment is established in the station based on the distribution of regular scatterers in the environment. The path loss is initially calculated using ray tracing (RT) technology, and the error is calculated based on the measured path loss. Then, based on the distribution of non-regular scatterers in the environment, environmental feature vectors such as corners and shading depth are extracted, and the radial basis function (RBF) neural network is used to predict the calculation error of path loss. The impact of irregular scatterers on signal propagation is calculated, and the accurate prediction of path loss in the substation is realized. Finally, the performance of the proposed model is trained and verified based on the measured data at 2.4 GHz in the substation. Simulation results show that this method reduces the Root Mean Square Error (RMSE) by about 5.28 dB compared with the traditional channel modeling method. This effectively improves the prediction accuracy of the wireless channel path loss in complex environments such as substations.