An Offline Parameter Self-Learning Method Considering Inverter Nonlinearity With Zero-Axis Voltage
Qiwei Wang, Nannan Zhao, Gaolin Wang, Shouhua Zhao, Zhixue Chen, Guoqiang Zhang, Dianguo Xu
Abstract
In the voltage source inverter applications, inverter nonlinearities would affect the parameter identification process in many ways. Hence, this article proposes an offline identification method for resistance and <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">dq</i> -axis inductance surface by considering the inverter nonlinearity characteristics. A variable amplitude square-wave injection (VASI) scheme is proposed for the <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">dq</i> -axis inductance identification. The VASI method achieves the inductance identification with a novel data sampling strategy. Meanwhile, it can also establish the inductance surfaces by only a few identified data points with a polynomial fitting algorithm, which greatly reduces the identification time compared with the existing methods. The resistance identification is realized by a slope signal injection method, in which the effect of IGBT voltage drop is analyzed. In order to improve the identification accuracy, the inverter nonlinearities are compensated by a self-learning method considering the zero-axis voltage at different rotor positions. At the same time, the sampling error in zero current zones of abc-phases is researched. In order to verify the effectiveness and generality, the proposed method is carried out on two different test machines and confirmed by finite element analysis.