摘  要：矢量控制系统中的转子磁场定向的精度取决于异步电动机的参数。在传统的利用空载、短路实验测量电机参数方法的基础上，本文提出一种利用变频器自身的资源，不需要外接电路的方案来对电机参数进行辨识。为了克服由于数据采样精度所照成的参数波动较大的缺点，提高所辨识参数的精度，利用上位机中基于自组织竞争神经网络算法对所得参数组进行学习以获得最优参数。试验结果表明：该神经网络可快速收敛，并可抑制参数异常点的干扰，获得较为精确的参数。
关键词： 异步电动机；参数辨识；自组织竞争；神经网络

Abstract: The precision of rotor flux orientation is relied on the precision of motor parameter in vector control system. A new method to measure parameters of motor used resources of the transducer has been achieved without any external circuit. The achieved parameters were studied by self-organized competition neural network to overcome the large parameter fluctuation range caused by data sampling accuracy and improve the accuracy of themselves. The experimental results indicate that this network icon verges quickly, suppresses the interference of outliers and gets exact parameters.
Key words: asynchronous motor; parameter estimation; self-organize competition; neural network