Abstract: To raise the level of hydropower consumption during flood season, an objective of minimizing abandoned hydropower cost is added to the previous minimum cost model, and a short-term power balance model that takes this cost into account is developed for the power system. And an improved particle swarm optimization algorithm is formulated based on a rolling correction strategy to overcome the difficulty in obtaining effective feasible solutions satisfying large-scale complex constraints and to avoid premature convergence and local optimization. Calculations of the Dianzhong regional power grid in Yunnan Province shows that relative to the minimum cost model, although the cost of the other energy sources is increased by 1.42 million Yuan, the abandoned hydropower in flood season is decreased by 7.14 million kWh and the total system cost by 7.0%. Relative to the minimum abandoned hydropower model, although the abandoned hydropower is increased by 2.85 million kWh, the thermal power cost is decreased by 17.95 million yuan and the total system cost by 21.7%. In the period of peak regulation capacity lacking, a proper arrangement of the other power stations for peak regulation can effectively reduce the abandoned hydropower and total cost.

Key words: power system, short-term power balance model, abandoned hydropower, particle swarm optimization algorithm, peak regulation