Abstract: Urban rail transit substation equipment is the core hub of urban rail transit power supply system, and its operating status directly determines the safety and stability of train power supply. In response to the pain points of traditional substation equipment monitoring methods being single, passive safety control, and lagging fault disposal, this article integrates IoT perception, artificial intelligence, big language models, and deep learning technology to carry out practical research on intelligent monitoring and safety control of urban rail substation equipment. By constructing an integrated system of multi-dimensional perception, intelligent analysis, and collaborative control, real-time monitoring of equipment status, advanced warning of hidden dangers, accurate fault diagnosis, and safety closed-loop control can be achieved, effectively improving power supply reliability, reducing operation and maintenance costs, and providing practical reference for the transformation of urban rail substation equipment from periodic maintenance to predictive maintenance.