Distributed photovoltaic (PV) is of great significance to realizing the objectives of “carbon peaking and carbon neutrality” and establishing a new power system mainly supported by new energy sources. In this study, the optimization operation of the household PV-energy storage system under the present step-peak valley tariff mechanism was investigated. Firstly, the structure of household PV-energy storage system and tariff mechanism were introduced. Secondly, by considering the impact of step-peak valley tariff on users’ energy use strategy on long time scale, a two-layer rolling optimization operation strategy of household PV-energy storage system based on model predictive control (MPC) was proposed, where the upper-layer model focused on step tariff-based annual rolling optimization with maximum annual comprehensive revenue as objective; in the lower-layer model, the operation scheme of PV and energy storage was based on the optimization results of the upper layer, and the deviation of system state caused by uncertain factors was corrected. In addition, the Informer model was used to achieve accurate prediction of the system state on a long time scale. Finally, the simulation results show that the proposed strategy can delay the arrival of high step tariffs and effectively improve the overall revenue of residential users.