Intelligent distributed protection and control is a promising route towards flexible and safe operation of the medium-voltage distribution network with widespread access to distributed energy resources. A fundamental premise of distributed decision-making is that each smart terminal unit (STU) can identify the topology of the feeder and track its changes. A distributed topology identification algorithm with high fault tolerance based on peer-to-peer communication is proposed in this paper. The STU configured at the looped network node requires no pre-set static topology of the feeder group,but only relies on local measurement and information exchange with neighboring STUs. Through rule judgment,it can realize the dynamic tracking and monitoring of the complete topology and open-loop mode of the feeder group,which reduces the configuration requirements of the algorithm on the terminal. Meanwhile,the method only needs each STU to exchange the local measurement and judgment information,which reduces the requirement of communication. Test examples show that the proposed method is reasonable and effective,and it can significantly improve the tolerance of topological identification to measurement errors by combining local measurement mutual checking and contralateral connectivity predictive correction.