The uncertainty of electricity consumption by power customers brings great challenges to the treatment of three-phase load imbalance in distribution station areas. The ubiquitous power internet of things is an effective technical means to control three-phase load imbalance from the source. A three-phase imbalance control method based on dynamic load programming is proposed. First, the accurate watt-hour meter-phase correlation is obtained by identifying the watt-hour meter-phase correlation by HPLC and comparing the changing trend of the watt-hour meter voltage sequence. Secondly, the curve of the user load characteristic is established using the high frequency load data of a watt-hour meter. Finally, the historical load data of single-phase users are correlated and analyzed, and the load dynamic programming algorithm is used to get a suggestion for adjustment of the watt-hour meter-phase relationship of the household meter. The practical application results of examples show that the adjustment accuracy of this method is better than that of the power distribution line load switching method and the three-phase imbalance phase-changing switch managing method. This work is supported by Science and Technology Planning Program of Sichuan Province (No. 2016GZ0143).