基于神经网络算法的可控电抗器谐波特性研究  

Study on Harmonic Characteristics of Controllable Reactor Based on Neural Network Algorithm

作　　者：程汉湘 熊培枫 夏健新 CHENG Hanxiang;XIONG Peifeng;XIA Jianxin(Tianhe College of Guangdong Polytechnic Normal University,Guangzhou,Guangdong 510540,China)

机构地区：[1]广东技术师范大学天河学院,广东广州510540

出　　处：《广东电力》2019年第4期34-40,共7页Guangdong Electric Power

基　　金：广东省重点培育学科电气工程粤教研函[2017]1号文

摘　　要：为改善磁阀式可控电抗器(magnetically controlled reactor,MCR)的动态控制过程,采用基于神经网络辨识的偏磁轨迹计算方法对可控电抗器谐波特性进行了研究。首先对比晶闸管控制电抗器(thyristor controlled reactor,TCR)和MCR的工作原理和性能参数,对于两者所产生的谐波幅值进行理论分析;然后以试验数据作为训练样本,运用神经网络模型的方法,分析MCR的谐波特点;再根据同一容量和电压等级的TCR和MCR,比较它们之间的谐波变化规律,据此指出减少MCR谐波的具体方法,同时提出提高MCR的响应速度的方法。这些方法提高了MCR动态控制过程中的稳定性和可靠性,其成果已经在10 kV和35 kV的变电站中得到应用。To improve dynamic control process of the magnetically controlled reactor (MCR),this paper uses a calculation method of magnetic bias trajectory based on the neural network identification for study.It firstly compares working principles and performance parameters of the thyristor controlled reactor (TCR) and the MCR and analyzes harmonic amplitudes of these two kinds of reactors.Then taking experimental data as training samples,it applies the method of neural network model to analyze harmonic characteristic of the MCR.By comparing variation laws of harmonic of the TCR and the MCR at the same level of capacity and voltage,it points out specific method for reducing harmonic of the MCR and improving the response speed at the same time.These methods improve stability and reliability in the process of dynamic control of the MCR and corresponding achievements have been applied in 10 kV and 35 kV substations.

关 键 词：可控电抗器 磁阀式可控电抗器 晶闸管控制电抗器 谐波 人工神经网络 

分 类 号：TM47[电气工程—电器]