With the increasing penetration of renewable energy, the issue of insufficient grid flexibility has become more prominent. As a new approach for integrating and managing distributed resources, virtual power plants (VPP) effectively enhance system flexibility by enabling multi-energy complementarity on the supply side and flexible interaction on the demand side. Considering the output uncertainty of distributed wind and solar resources, this paper proposes a VPP aggregation cost risk decision-making method based on stochastic dominance theory. By introducing second-order stochastic dominance constraints to manage aggregation cost risks, the method effectively addresses the limitations of traditional risk measurement models, which often rely heavily on distributional assumptions and lack flexibility. Based on the conditional value-at-risk (CVaR) risk management model, the feasible region for the second-order stochastic dominance constraints is clarified. On this basis, a multi-index, single-scenario benchmark variable determination method that integrates the Sharpe ratio and CVaR is proposed, providing guidance for risk-informed decision-making on the aggregation and regulation costs of heterogeneous distributed resources. Finally, simulation analysis on an 11-node distribution system verifies the effectiveness and superiority of the proposed risk decision-making method in VPP resource scheduling.