Abstract: In response to the complex operating environment of transmission lines, the diverse types of hidden dangers, and the poor real-time performance and low recognition accuracy of traditional monitoring technologies, this paper designs an online monitoring technology for transmission line hidden dangers based on intelligent perception collection and long short- term memory networks. This article designs an intelligent sensing acquisition terminal that integrates multiple sensors, achieving real-time synchronous acquisition and edge preprocessing of multi-source state data. At the same time, the collected temporal feature data is input into a long short-term memory network, realizing the learning of dynamic evolution rules of normal and hidden danger states of the line, and constructing a hidden danger recognition and early warning model. The experimental results show that the technology proposed in this article can effectively capture weak hidden danger signs, has high warning accuracy, and improves real-time operation and robustness, providing a feasible technical path for online intelligent and accurate monitoring of transmission line hidden dangers.