Abstract: The aim of this paper is to address the high-concurrency access problem of smart devices in new-type power systems. First, we construct a cloud-edge-end collaboration-based high-concurrency access management framework of smart devices and build a model for high-concurrency access management optimization problem with the goal of maximizing the average total load balance degree. Subsequently, a pre-allocation and dynamic load balancing-based multi-scale high-concurrency access management algorithm is proposed to optimize the large-timescale access channel pre-allocation based on the edge-side queue state and device-side access success rate, as well as small-timescale load balance through dynamic bidding and load unbalance awareness. Finally, the proposed performance advantages of the proposed algorithms are confirmed through simulation analysis. The research content of this paper can improve high-concurrency access management performance, thereby enhancing the operational efficiency of power services.