中文摘要：

针对云粒子群算法（CPSO）在电力系统无功优化中易陷入局部极值和存在早熟收敛问题,将基于云数字特征（期望值、熵值、超熵值）编码的云粒子群算法进行了改进：依据解空间的变换将局部搜索和全局搜索相结合,用正态云算子实现粒子的进化学习和交叉变异操作.改进的算法在时间、存储量性能上有了明显的提高,将改进后的算法应用到IEEE30节点标准测试系统和玉门电网进行仿真运算,与其它算法进行比较.其结果表明：该方法在电力系统无功优化中能取得更好的全局最优解,加快了收敛速度,提高了收敛精度.

英文摘要：

Since the cloud particle swarm optimization is easily trapped in local minimum value and slow in convergence in reactive power optimization of the power system,the cloud particle swarm optimization is improved based on cloud digital features（Ex,En,He）,that is,the local search and the global search are combined according to the space transform,and the normal cloud particle is applied to the evolution of the learning process and the variation operation,so as to shorten the time and enlarge the storage of the improved algorithm.The improved algorithm is simulated to IEEE30 bus system and Yumen Power Grid,and it is compared with other algorithms.The result indicates that it can obtain much better global solution,accelerate the convergence speed and improve the convergence accuracy in the reactive power optimization of the power system.