The time-of-use tariff policy is mostly implemented with a fixed time slot delineation scheme, but with the large access of renewable energy on the supply side and the enhanced flexibility of broad load on the demand side, the relationship between supply and demand shows dynamic changes. Therefore, in order to study the optimization of peak and valley time slots of dynamic time-of-use tariffs under different time scales, this paper adopts the improved fuzzy-C means (FCM) clustering algorithm to construct peak and valley time slot delineation model based on the comprehensive consideration of the large proportion of renewable energy connected to the networks and the demand-side response. The model determines the optimal time slot delineation results through the clustering analysis of the daily change curves of the net load under different time scales. According to the example analysis of Liaoning Province’s net load data from April 2021 to March 2022, renewable energy has anti-peaking characteristics, which causes the peak-to-valley difference of the net load curve to increase, and users’ response to the time slot delineation of the time-of-use tariff policy has a lag and timeliness. Therefore, it is recommended to dynamically adjust the time slot delineation of the time-of-peak tariff policy every 3?4 months to better tap the potential of users’ demand side response and promote peak shaving and valley filling. However, if the peak and valley time slots are frequently adjusted, it will be difficult for users to adjust their electricity consumption in time to respond.