An operation optimization model minimizing the thermal cost for short-term hydrothermal system with
cascaded reservoirs is established in the paper. Several constraints, such as power balance, water balance and the physical
characteristics of the plants, are considered. To overcome the shortcomings of a standard particle swarm optimizer(PSO),
a modified PSO(MPSO) is developed to solve this scheduling problem. Mutation and migration operators are introduced
respectively in the proposed method. The mutation operator makes particles learn others’ personal best solution with a
dynamically decrease probability. The migration operator regenerates a newly diverse population of individuals, once the
crowding level is smaller than the desired tolerance of population diversity. In order to inspect and verify the reasonable
of the operation optimization model and the validity of the proposed method, a famous hydrothermal power system in the
literatures is tested. The simulation results show that MPSO is more effective than other algorithms.