中文摘要：

粒子群优化算法是通过粒子记忆、追随当前最优粒子，并不断更新自己的位置和速度来寻找问题的最优解。为了克服标准粒子群算法存在着早熟收敛、难以处理问题约束条件等缺点，本研究对递减惯性权值进行了改进，将其表示为粒子群进化速度与群体平均适应度方差的函数；给出了适合PSO算法的约束处理机制，提出了一种改进自适应粒子群算法，并将其应用于水库优化调度中。实例计算并与经典方法相比，表明该方法原理简单、易编程实现，能以较快的速度收敛于全局最优解。

英文摘要：

Particle swarm optimizer（PSO） searches the best solution of a problem by remembering and following the excellent particle and updating own position and speed continuously. In order to overcome the defect of premature and difficulty of dealing with constraint, this paper presents a modified adaptive PSO （MAPSO） which expresses inertia weight in a function determined by the evolution speed and the fitness variance of particle swarms and proposes a constraint handling strategy suit for PSO. A hydropower station operation demonstrates the successful application of the modified adaptive PSO. Study results show that the MAPSO is a simple, programming easily optimal algorithm and can find the global optimum solution quickly compared with traditional method.