This paper proposes an optimization scheduling model for power systems taking into account the flexibility of source load side regulation resource given the characteristics of peak shaving and the problem of new energy consumption faced by the power grids. First, an improved SSE-PFCM clustering algorithm is used to calculate the typical day of the load based on the characteristics of the load. Secondly, based on the different response behaviors of load users to electricity price changes, two types of loads with demand response potential are extracted. Fuzzy load transfer rate models and demand price elasticity models considering optimistic response membership degrees are constructed. Then, taking into account the demand response uncertainty model and the deep peak shaving of thermal power units, an optimal scheduling model is constructed with the objective of minimizing the overall operating cost of the system. An improved particle swarm optimization algorithm with the introduction of step size elasticity coefficient is used to analyze the problem. Finally, taking the improved IEEE 30-bus system as an example, simulation calculations are conducted with various scenarios, and the results show that the proposed strategy can effectively improve the renewable energy consumption capacity of the system and the economy of system operation.