【作者】 王萍；

【导师】 孙雨耕；

【作者基本信息】 天津大学 ， 电工理论与新技术， 2005， 博士

【摘要】 有源电力滤波器(Active Power Filter——APF)可用来对电力系统中的谐波、无功等进行综合补偿,是解决电力谐波问题最具前景的手段之一。有源电力滤波器中畸变电流的检测与控制是决定其补偿效果的关键环节,是研究APF应用的基础。由于测控对象是个非线性的复杂系统,因而智能测控技术的应用具有很大潜力。本文研究有源电力滤波器畸变电流的智能测控方法,以适应电力系统的复杂性及不同有源电力滤波器应用的要求。本文将Hopfield神经网络优化理论用于谐波检测中,提出了一种基于Hopfield神经网络(HNN)的谐波电流检测方法。该方法的优点是能够自适应地检测出电网基波和各次谐波分量的幅值和相角,采样数据实时处理,速度快,实时性好,网络不需要预先进行训练。采用一种新的基于正交锁相环(QPLL)的电网频率检测方法,并与Hopfield神经网络相结合,构成QPLL-HNN谐波电流检测系统。该系统能够同时适用于已知电网精确频率和电网基频发生变化的场合,实现电网频率及各次谐波的精确检测。仿真结果表明检测方法具有很好的实时性和较高的检测精度以及自适应跟踪负载电流变化的能力。提出有源电力滤波器补偿电流的智能滞环控制方法,包括基于遗传算法的滞环控制方法和基于神经网络的滞环控制方法。遗传算法控制能够优化开关时刻,进而实现补偿电流的优化控制。该方法的主要优点是可以有效降低有源电力滤波器的开关频率,进而减小滤波器工作损耗。用BP神经网络构成滞环控制器,并通过DSP数字实现。该方案可实现神经网络对快速变量的控制,提高滞环控制的性能,使系统对参数的变化有较强的不灵敏性和鲁棒性。仿真和实验结果证明控制策略正确、可行。本文将并联电流型有源电力滤波器加在逆变器直流侧,取代电解电容,应用模糊逻辑控制和滞环控制实现电流型APF的控制,使有源电力滤波器补偿逆变器产生的谐波。理论分析和仿真、实验研究证明,有源电力滤波器可以有效补偿逆变器输入谐波电流,从而“硅解”电解电容,实现逆变器直流侧的有源补偿。更多还原

【Abstract】 Active Power Filter (APF) is one of the most promising methods to solve theharmonic problem, which can be used for integrated compensating harmonic,reactive currents. The detection and control methods of distortion currents for APFare key factors affecting not only the compensation performance of APF, but also theextension application. Artificial intelligence techniques have become essential to thedetection and control for distortion currents of power system. The thesis studies newintelligent detection and control approaches of distortion currents for active powerfilters to satisfy the diverse requirements for APF to apply in the modern powersystem.An adaptive detecting approach of the harmonic currents based on Hopfieldneural network (HNN) optimum theory is proposed in this paper. Instead of trainingthe network, the basic principle of neural network is applied to determine adaptivelythe magnitude and phase of the fundamental and each of the harmonic currentcomponents in real time. A new method for estimation of power frequency based onquadrature phase-locked loop (QPLL) concept is used to Hopfield neural network.The QPLL--HNN harmonic estimation system can be applied to the precisefrequency and each of the harmonics for power system. The computer simulationsresults show that the methods have some advantages, such as real time, highaccuracy, adaptability to load currents, etc.The new intelligent hysteresis control methods based on Genetic Algorithm andneural network to compensate currents for active power filters are proposed in thepaper. Genetic algorithm (GA) is employed to optimize the switching instants inmain circuit to minimize total harmonic distortion in the power system as well asdecrease operating frequency of APF greatly. A hysteresis controller is realized by aback propagation neural network and the digital signal processor. The controller hasa high performance in response of current and robustness and insensitivity to theparameter changes. The simulations and experiments results prove the effectivenessof the control strategy.A shunt active power filter (APF) as harmonic generator taken to the dc link ofthe inverter eliminates harmonic wave generated by the inverter, so implement anew type of electrolytic capacitorless inverter circuit with harmonic compensation inthis paper. Control is mixed by Fuzzy logic control and hysteresis control, whichmade APF generate the harmonics current compensated accurately. The system isinvestigated analytically as well as experimentally. The obtained results arepresented to prove the correctness of the proposed method.更多还原