正弦逆变波形的神经网络内膜控制算法  

Output Waveform Control of Single-Phase Inverter Based on Neural Network Internal Model Theory

作　　者：杨金辉[1] 郜克存[2] 戴瑜兴[3,2] 洪天宇 

机构地区：[1]华为技术有限公司,北京100095 [2]湖南大学电气与信息工程学院,湖南长沙410082 [3]温州大学物理与电子信息工程学院,浙江温州325035 [4]斯伦贝谢中国公司,北京100015

出　　处：《电子学报》2012年第7期1345-1350,共6页Acta Electronica Sinica

基　　金：住房和城乡建设部科学技术研究开发项目(No.2008-k6-22)

摘　　要：单相逆变电源的输出电压波形质量是衡量其性能的重要指标之一.本文提出了一种正弦逆变波形的神经网络内膜控制算法,建立两个BP神经网络预估器,一个作为单相逆变器的内部模型,预测实际的波形输出;一个根据预测误差建立内模控制器,在线修正和补偿内部模型使之最大程度的匹配单相逆变器.仿真和实验结果表明,该算法克服了系统中存在的不确定性,有效的提高了系统的逆变波形质量和负载适应性.The quality of output waveform is one important factor for the single-phase inverter. This paper proposed the algo- rithm based on neural network internal model theory, which is used for an output sine waveform control. Based on this control algo- rithm,two back propagation estimate neural networks were established. One is the internal model of single-phase inverter, which is used to estimate the actual output waveform. The other one is used to make the internal model to fit the actual single-phase inverter due to the estimate en＇or. We finished the simulation and experiment, where the algorithm was proved that it could improve the out- put waveform quality and load compatibility.

关 键 词：单相逆变电源 内模原理 BP神经网络 波形控制 

分 类 号：TM921.5[电气工程—电力电子与电力传动]