基于ANN的三电平逆变器开关矢量选择器

Output vector selector based on ANN for a three-level inverter

大容量传动场合中,多电平电压型逆变器得到广泛的应用,但是多电平逆变电路输出矢量的选择较为困难。在细致地分析了三电平电路的输出矢量后,总结了其矢量变化的规律,以异步电动机直接转矩控制为例,提出一种根据磁链和转矩误差选择输出矢量变化量的方法,简化了输出电压矢量的选择过程。并利用所提的方式设计输出矢量选择表,以其为导师对人工神经元网络进行训练,实现了三电平逆变器直接转矩控制的人工神经元网络输出矢量选择器。这种开关矢量选择器可以极大地简化获得输出电压矢量的过程,并可推广至更多电平数的逆变电路中。最后,通过仿真实验验证了所提出矢量选择器的有效性。

Multilevel Voltage Source Inverters (VSI) are widely used for high power drive systems, but output vector selection is difficult. An output vector shifting method was developed based on the output vector characteristics of a threelevel inverter. A simplified method was then used to choose the output vector for Direct Torque Control (DTC) of a threelevel inverter fed induction motor. The output variance other than the output vector was selected according to the error in the stator flux and the torque, so the output vector selection for a multilevel inverter is similar to that for a twolevel inverter except for the effect of the original state. An output vector table produced using this method was used to train a switching vector selector using an Artificial Neural Network (ANN). The selector could then quickly select an output vector which could be applied to inverters having more than three levels.