Deterministic trend currents and static reconstruction methods cannot account for the influence of uncertainties in the power grid. For this reason, a multi-objective dynamic reconstruction method based on load and DG randomness is proposed. Firstly, the probability model of DG and load is established, and the semi-invariant and Gram-Charlier series are used to calculate the probabilistic load flow. Then, a SA-PAM algorithm is proposed to perform load clustering to realize the reconstruction period division. Subsequently, a distribution network reconstruction model with network loss expectation, voltage over-limit probability and voltage offset is established. In order to improve the speed and efficiency of the model, a multi-group parallel immune algorithm is proposed to achieve fast convergence and global search. Finally, the effectiveness of the proposed method is verified by simulation of IEEE33 node system.
