Abstract: To ensure the reliable integration of distributed photovoltaic clusters, this study proposes a shared energy storage capacity allocation method that takes into account the uncertainty of photovoltaic power prediction errors. First,the distributed photovoltaic output is predicted based on the attention mechanism and the long-term short-term memory neural network（LSTM）,and then the prediction error is obtained by comparing it with the actual value of photovoltaic output under different typical meteorological conditions.Second,with the optimal cost as the goal,a shared energy storage capacity allocation model is established to track the deviation of photovoltaic output plan,and the uncertainty of prediction error is described by introducing robust opportunity planning constraints,and the convex approximation method is used to transform the original model into a deterministic model for solving. Finally, the simulation results show that the proposed method can maximize the economy of energy storage configuration while ensuring the compensation effect.