Extreme harsh environments such as strong electromagnetic waves and blizzards in densely populated transmission channels in remote areas can have a serious impact on wireless channels, leading to decreased signal quality, damaged link stability, and even communication interruptions. To ensure reliable and stable data transmission for online monitoring, mobile inspection, on-site operations, and other businesses in dense transmission channels, this paper proposes a wireless channel parameter extraction method based on deep reinforcement learning, and designs a wireless channel prediction model based on an improved neural network. By using artificial intelligence methods to accurately estimate and predict wireless channel characteristics, the propagation mechanism and channel characteristics of radio waves in complex power environments are obtained, providing data support for the optimized deployment of the Shagehuang wireless communication system and effective guarantee for reliable communication in power transmission businesses.