Abstract: A high proportion of renewable energy in power systems puts higher requirements for peak shaving. The correlation between distributed energy resources and flexible loads is vital to virtual power plant planning. This paper contributes an optimal resources aggregation model of a virtual power plant considering multi-dimensional generation-load correlation and peak shaving ability. First, the spatio-temporal correlation between generation and load is modeled through the R-vine Copula embedding Kshape algorithm to reduce high-dimensional joint scenarios. Also, a synthetical indicator termed multi-dimensional peak shaving ability is proposed to quantify the overall responsiveness and credibility of the aggregated resources and differentiate the contribution of each. Finally, A multi-objective VPP planning model is proposed to promote renewable energy consumption and provide regulation capacity for the power grid to determine the optimal resource aggregation strategy. The performance of the proposed indicator and planning method is demonstrated through a practical application to a district of Langfang City, Hebei province, obtaining conclusions with certain guiding significance.