Objective In order to describe the impact of the uncertainty of distributed photovoltaic (DPV) output and load on optimal configuration of DPV, an optimal configuration interval model of distributed photovoltaic was proposed.

Method Firstly, the adaptive interval type-2 fuzzy C-Means clustering algorithm was employed to establish the interval clustering of DPV output and load under multiple scenarios. Then, taking the comprehensive cost minimum as the objective function, chance constrained method was used to deal with reverse power flow constraint and state variables constraints. An interval model based on interval clustering for DPV optimal configuration was established, and the constructed interval model was transformed into up and low bound sub-models. Finally, the firefly algorithm was improved by adopting the mix-level orthogonal initial population experimental and introducing the distance regulation random coefficient and applied it to solve the up and low bound sub-models. In addition, for the characteristics of the DPV output, based on correlation coefficient between DPV output and load demand, a method suitable for determining candidate bus for DPV was proposed to further reduce the computation complexity of the model.

Result A modified IEEE 33-bus system has verified the effectiveness of the proposed model and the algorithm.

Conclusion The research in this paper can provide a strong theoretical basis for the optimal configuration of DPV considering the uncertainty of DPV output and load.