摘要
统一电能质量控制器可同时补偿电网畸变电压和抑制负载谐波电流。为此,构造了一种基于反向传播算法的三层前馈神经网络用来检测并联型有源电力滤波器的谐波电流,离线训练收敛后实现在线功能,对串联型有源电力滤波器谐波电压检测采用了畸变电压参考量比较检测方法;建立了统一电能质量控制器的系统仿真模型,利用其对各种电能质量问题的补偿性能进行了仿真研究,并对补偿前后负载和电源电流/电压进行了频谱分析。研究结果表明,统一电能质量控制器集电压补偿、电流补偿于一体,可有效实现多重电能质量调节功能。
Unified power quality conditioner （UPQC）, known as universal power quality conditioner, can mitigate the power supply distortion and eliminate the load harmonic current at the same time in power systems. A three-layer feedforward neural network based on back-propagation algorithm used to detect harmonic of parallel active power filter on-line after converged network in off-line way, is constructed in this paper. The distortion voltage reference compare method for detecting harmonic voltage of series active power filter is presented. The UPQC model is built using MATLAB to study the compensation performance of UPQC. Spectrum is ana- lyzed before and after compensation for current/voltage of load and supply respectively. Simulation results demonstrate that UPQC combines the functions of a voltage compensator and a current compensator, which can effectively regulate multiple power quality problem.
出处
《电力系统及其自动化学报》
CSCD
北大核心
2009年第4期30-35,共6页
Proceedings of the CSU-EPSA
基金
内蒙古自治区自然科学基金重大项目(200711020801)
内蒙古自治区自然科学基金项目(200607010809)
内蒙古自治区高等学校科学研究项目(NJ05050)
内蒙古自治区自然科学基金项目(20080404MS0907)
关键词
统一电能质量控制器
多层前馈神经网络
建模
仿真
电能质量
unified power quality conditioner
multilayer feedforward neural network（MLFNN）
modeling
simulation
power quality
