摘要
针对无功优化中确定无功补偿点和无功补偿容量的问题,本文提出了一种基于奇异值分解理论和多策略自适应粒子群优化算法(MS-APSO)的无功优化法。首先基于潮流计算中的雅可比矩阵奇异值分解以确定电压稳定性较弱的节点作为无功补偿节点;然后以线路有功损耗、负荷节点电压偏差最小以及节点稳定度最大为目标优化无功补偿量。为解决迭代后期算法收敛速度降低、粒子群多样性下降等问题,提出了多策略自适应改进算法以寻求全局最优解,综合考虑了粒子群多样性、惯性权重、越限重置和变异的影响,有效提高了算法前期的收敛速度和后期的寻优能力。最后,改进算法的有效性在IEEE 118算例中得到了验证。结果表明,改进后算法降损率与传统方法相比可以提高38.6%。
In view of such issue as determining reactive power compensation point and reactive power compensation capacity in the reactive power compen s ation,in this paper,a kind of re active power optimiz ation method based on singular value decomposition and multi-strategy adaptive particle swarm optimization(IS-APSO) algorithm is proposed.Firstly,the weak node of voltage stability is taken as reactive power node based on the singular value decomposition in the power flow c alculation;then the reactive power compensation is optimized with minimum active power loss of line and node voltage deviation and maximum node stability as target.For solving such problem as reduction of convergence speed of algorithm and reduction of diversity of particle swarm in the later period of the optimization,MS-APSO algorithmis proposed to effectively seek optimal solution.The influence of diversity of particle swam,inertia weight,over-limit reset and mutation is considered comprehensively,which improves effectively the convergence rate of the algorithm at earlier stage and optimization ability at later stage.Finally,the effectiveness of the improved algorithm is verified in IEEE 118 calculation.It is shown by the result that the loss reduction rate of algorithm after improvement,compared to the traditional method,can improve by 38.6%.
作者
陈春萌
梁英
张舒捷
CHEN Chunmeng;LIANG Ying;ZHANG Shujie(State Grid Qinghai Electric Power Corporation,Xining 810000,China;State Grid Qinghai Electric Power Research Institute,Xining 810000,China)
出处
《电力电容器与无功补偿》
北大核心
2020年第4期102-108,共7页
Power Capacitor & Reactive Power Compensation
基金
国网青海省电力公司科技项目(青海主网高损线路分析及降损措施研究)。
关键词
电力系统
无功优化
奇异值分解
粒子群优化算法(PSO)
多策略
自适应
power system
reactive power optimization
singular value decomposition
particle swarm optimization(PSO)
multi-strategy
self-adaptive