Because of the large number of line sections and branch switches in a distribution network, it is difficult to coordinate the protection level difference of the network, and fault removal is fuzzy, and the fault range is expanded. The application of a distribution automation system can solve this problem and complete the self-healing of the network. At present, distribution automation is mainly centralized by the master station, that is, the information on electrical and switching quantity of each node is collected and sent to the master station. The master station completes logic judgment and issues strategies to realize fault isolation and power grid self-healing. There are many data transmission and verification links in the master station mode, and the fault isolation time is usually at the minute level. When the distribution network is self-healing, the main power supply line needs to be powered off for a short time. In this paper, distributed feeder automation based on 5G communication technology is adopted, and millisecond level fault local isolation and network reconstruction are realized through sinking calculation and logical discrimination between nodes and distribution terminals, with accurate fault removal. The explored optimization scheme of distribution network protection setting realizes fault isolation and distribution network reconstruction when the main power supply line is not powered off, and achieves the "zero perception" of power failure by users through the integration of distribution network protection setting and distributed feeder automation fault discrimination strategy. The project has been implemented and applied in the Chengdu urban distribution network for the first time, achieving good results, effectively solving the problem of difficult coordination of protection level difference in urban areas, and accumulating experience for the expansion and application of 5G communication protection in distribution networks and other voltage level power grids. This work is supported by the National Natural Science Foundation of China (No. 52190421018Q).