Abstract: With the development of 5G-based power internet of things, various new services are emerging, such as electricity consumption information collection, power transmission and transformation status monitoring, and precise load control. An intelligent network slicing management method is proposed for the URLLC (ultra reliable low latency communication) scenario of the power internet of things. This method integrates 5G slicing and MEC (mobile edge computing) technologies to establish the delay, energy consumption and reliability models of power service, and the slicing resources are optimized through the DQN (deep Q network) algorithm. Simulation experiments show that the reliability of the intelligent network slicing management method proposed in the study is higher than 98%, and it is better than the classic coordinate block descent method and the average resource allocation method.