With the promotion of "Internet +" construction, as an important application of power Internet of Things （IoT） construction, multi-station integration needs to study its modular configuration scheme suitable for different regions. Considering the life cycle attenuation of energy storage and the economic requirements of multi-station integration operators, this paper proposes a bilevel stochastic programming model considering the operation characteristics of each functional substation. Firstly, the output or load model under normal operation is determined according to the operation characteristics of functional substations that need to be fused. Secondly, considering the uncertainty of source load during long-term operation, the corresponding uncertainty characterization and processing method are adopted according to the load characteristics. The two-layer stochastic optimization model is used to solve the optimal configuration model in a long-time scale, and the operation results in the whole life cycle are equivalent to the expected values of several typical operation scenarios. Finally, the model is validated in a typical 220 kV substation scenario.