Abstract: To improve the flexible regulation ability of renewable energy stations, a distributed robust optimization method based on Bayesian theory is proposed for the allocation of energy storage with stable support demands. To scientifically describe the randomness of renewable energy outputs and accurately characterize the frequency modulation demand caused by large-scale renewable energy grid connections, considering the uncertainty of probability distribution of the uncertain variables, based on the robust optimization of distribution, the Bayesian multivariate nonlinear model and the Bayesian mixed-effect model are proposed to construct the uncertain set of wind power and PV respectively. This method will make up for the defects of the fuzzy set subjectively constructed and the representation of uncertainty only under a single time section. Finally, a case study is given to verify the effectiveness and feasibility of the proposed method in the energy storage configuration. And the sensitivity of energy storage allocation results is quantitatively analyzed under the influence of power generation uncertainty. The method proposed in this paper provides theoretical guidance for energy storage allocation for large-scale renewable energy stations to meet their inertia support and frequency modulation requirements.