The continuous growth of loads such as electric vehicles（EVs）and air conditioners（ACs）has made transformer overload and accelerated aging problems more and more serious.An optimal coordination method for residential loads that dynamically determines the maximum load is proposed based on the hot-spot temperature（HST）of the transformer winding. In order to meet the scale requirements of loads, the duality principle is used to decompose the original problem into a distributed optimization problem of each flexible load. The simulation case considers flexible loads such as EVs and ACs, and compares four schemes：dynamic congestion management, static congestion management, self-interested mode, and free power consumption. Results show that this method can effectively implement congestion management and control the HST while significantly reducing electricity costs.