Multi-stage distributionally robust optimization with risk aversion

Two-stage risk-neutral stochastic optimization problem has been widely studied recently. The goals of our research are to construct a two-stage distributionally robust model with risk aversion and extend it multi-stage case. We use coherent measure, Conditional Value-at-Risk, describe risk. Due the computational complexity nonlinear objective function proposed model, two decomposition methods based on cutting planes algorithm solve distributional problems, respectively. To verify validity models, we give applications multi-product assembly portfolio selection problem, Compared models more robust.