Abstract: To improve the premature convergence problem of traditional shuffled frog leaping algorithm
(SFLA), normal cloud mutation operation is used to optimize the solution of each group by the cloud
model’s characteristics of uncertainty with certainty, stability, and flexibility in knowledge expression, and
the beat step size is adjusted for local depth search. Using this technique, we have developed a normal cloud
mutation-shuffled frog leaping algorithm (NCM-SFLA) that can avoid easy trapping into local optimum in
calculation of evolution. This new algorithm was applied to short-term optimal dispatch of cascade
hydropower stations. Application in a case study shows that it has better global search ability and faster
convergence speed than those of DPSA, SFLA or PSO and it is effective in solution of short-term optimal
operation of cascade hydropower stations.