Abstract: In this paper we propose a capacity configuration method based on flexible peak shaving transformation to address the issue of the large random fluctuations in the output of wind and photovoltaic (PV), which results in the imbalance between source and load. This method effectively suppresses wind and solar fluctuations. Firstly, a model is constructed, incorporating wind and PV output, deep peak shaving of thermal power units, as well as the state of charge(SOC) of energy storage battery. On this basis, considering the uncertainty of wind and PV power output and load demand, an annualized cost model is constructed for power generation and its state constraints is given which take minimizing the annualized cost of the system and the variance of new energy fluctuations as the objective functions. The uncertainty models are then established based on envelope constraint for wind power, PV and load intervals. Additionally, the equipment capacity configuration optimization strategies are presented based on IGDT prediction bias. The simulation verification is finally carried out by designing an example based on the actual data of a region in Hebei Province. From the comparative study on the improvement of indexes with those without flexibility transformation or deep peak shaving, it can be found that, with the flexibility of thermal power transformation and deep peak shaving taken into account, the system owns the advantages of paying more attention to new energy resources and better ensuring the balance of source and load as well as the capacity configuration. Thus, in the event of significant prediction bias, it can still meet the demands of the actual load.-solar-thermal-storage System Capacity Considering Flexible Reconstruction of Thermal Power and Deep Peak Shaving.