Self-organizing maps for scenario reduction in long-term hydropower scheduling

This paper presents a hybrid model framework consisting of a self-organizing map clustering and the Scenario Fan Simulator model for hydropower scheduling. First, the clustering model groups hydro inflow scenarios according to their features, such as trend, distance, and variation. Then, the prominent scenarios are selected from groups to feed into the hydropower scheduling model. The results show how the proposed model is able to conduct a disaggregated level of energy dispatching and deal with the multiple energy generators. Furthermore, the proposed hybrid model can significantly reduce the computational time by 80%, with the same load shedding indications compared to the case without scenario reduction.