摘要
为在线快速评估电力系统当前电压稳定性,建立了遗传算法优化的BP神经网络电压稳定评估模型。基于一般非线性方程极值分析原理,推导出系统静态电压稳定判据,由此提出评价系统静态电压稳定性的负荷法向阻抗模裕度指标(the normal impedance modulus margin index,NIMMI)。对比戴维南阻抗模裕度指标,法向阻抗模裕度指标的线性关系更好,适用于神经网络电压稳定在线预测。在系统同步功率扰动情形下,将系统薄弱区域节点的负荷有功、无功功率与节点的NIMMI值建立非线性映射关系,由此建立了以法向阻抗模裕度指标为样本的神经网络评估模型。用遗传算法优化BP神经网络的权值和阈值,提高了预测精度。Matlab仿真结果表明:相比传统潮流计算,法向阻抗模裕度指标的计算速度大大加快,更有利于实现系统电压稳定性的在线快速预测。
In order to evaluate power system voltage stability on-line rapidly, a BP neural network model optimized with genetic algorithm was proposed.Based on general principle of extreme value analysis of nonlinear equation, criterion of system static voltage stability was deduced; and normal impedance modulus margin index(NIMMI) suitable for power system voltage stability evaluation was proposed.Compared with Thevenin impedance modulus margin index, NIMMI linearity is better and more suitable for neural network prediction.Nonlinear mapping relationship of node active power(reactive power) and node NIMMI was established under system synchronous power disturbance.Genetic algorithm was used to optimize weights and threshold of BP neural network and improve prediction accuracy.Matlab simulation results show that NIMMI computation speed is greatly accelerated compared with traditional power flow calculation, so it is more favorable to implement on-line voltage stability evaluation.
出处
《电网技术》
EI
CSCD
北大核心
2016年第8期2389-2394,共6页
Power System Technology
基金
国家自然科学基金资助项目(51577053)~~
关键词
电力系统
电压稳定
法向动态等值阻抗
法向阻抗模裕度
遗传算法
BP神经网络
power system
voltage stability
normal dynamic equivalent impedance
normal impedance modulus margin index
genetic algorithm
BP neural network