Appendix_KMDR method.pdf

In addition to the risk-neutral SPDS shown in section Ⅲ, we formulate a two-stage mean-risk SPDS problem (MR- SPDS) handled by the proposed KMDR. The MR-SPDS cannot be directly solved by the commercial optimization solver, which requires reformulations considering the tractability. First, the original MR-SPDS problem is formulated in vectors and matrices. Next, we apply the Kullback-Leibler (KL) divergence-based LRA modelling to the CVaR-based mean-risk optimization framework.