基于BP神经网络磁滞模型的MCR偏磁特性研究

Study on the bias magnetic properties of MCR based on BP neural network hysteresis model

为了研究直流偏磁工作条件下磁阀式可控电抗器(MCR)的磁滞回线及输出电流的特性,根据MCR在直流偏磁工作条件下的实验数据,建立BP神经网络磁滞模型,得到磁滞回线。将MCR的电路与磁路耦合,得到绕组的输出电流。与实验波形对比,基本吻合。表明了该模型的正确性,为研究磁控电抗器的偏磁性能提供了有效方法。

To study magnetic valve type controlled reactor（ MCR） of the hysteresis loop and the characteristics of the output current under the working conditions of the dc bias magnetic,according to the experimental data of MCR under DC bias,the BP neural network hysteresis model is built to get the hysteresis loop. The circuit and magnetic circuit of MCR are coupled,thus the winding output current is obtained. Compared them with the experimental waveforms,the result is basic consistent. It shows the correctness of the model,and provides an effective method for the research of bias magnetic properties of magnetic control reactor.