基于改进遗传模拟退火算法的无功优化  

Research on Reactive Power Optimization Based on Improved Genetic Simulated Annealing Algorithm

作　　者：刘科研[1] 盛万兴[2] 李运华[1] 

机构地区：[1]北京航空航天大学自动化科学与电气工程学院,北京市海淀区100083 [2]中国电力科学研究院,北京市海淀区100085

出　　处：《电网技术》2007年第3期13-18,共6页Power System Technology

基　　金：国家重点基础研究发展计划项目(973项目)(G1998-030405)。~~

摘　　要：针对目前电力系统无功优化算法所存在的问题,提出了一种将遗传算法与模拟退火算法及牛顿下山法相结合的混合求解算法。首先根据个体适应度值进行自适应交叉和变异操作并采用模拟退火进行个体更新,以便增加群的多样性,避免陷入局部最优;然后采用牛顿下山法加快模拟退火部分的求解过程,并采用十进制整数编码和保存最优个体法来提高计算速度和精度。以IEEE30-bus系统和一某实际电力系统为例对所提出算法的性能和求解精度进行了测试,结果表明改进的混合遗传算法比传统的遗传算法在计算速度和全局收敛方面有了很大提高。To solve the problem existing in power system reactive power optimization algorithm, a new kind of hybrid genetic algorithm, which is composed of simple genetic algorithm, simulated annealing and downhill algorithm is proposed. At first according to the value of individual fitness the adaptive crossover and adaptive mutation are performed, thus the diversity of group is increased and the local optimum can be avoided. Then Newton downhill algorithm with a specifically probability is used to speed up the solving process of simulated annealing. To improve the computation speed and accuracy, decimal integer encoding and reserving the elitist are adopted. IEEE 30-bus system and an actual power system are used to test the performance of the proposed algorithm and its accuracy. Test results show that the improved hybrid genetic algorithm is much better than traditional genetic algorithm in both computation speed and global convergence.

关 键 词：无功优化 自适应遗传算法 遗传模拟退火算法 十进制整数编码 牛顿下山法 

分 类 号：TM744[电气工程—电力系统及自动化]