【目的】电压源逆变器的非线性特性引起输出电流畸变,进而导致电机输出转矩脉动,影响永磁同步电机驱动系统的控制精度和稳定性。为解决该问题,本文提出一种基于非线性干扰观测器(NDO)的在线补偿策略。【方法】首先定量分析逆变器死区时间、开关管压降等非线性因素;继而基于同步旋转坐标系电流方程构建NDO,设计饱和增益函数,并通过前馈补偿通道修正dq轴参考电压。然而,NDO对参数敏感性较高。因此,本文采用级联模型参考自适应参数辨识(CMRAPI)方案,先将级联模块参数辨识任务分解为定子电阻和永磁体磁链以及定子电感两组相互独立的辨识任务,每组任务通过模型参考自适应结构实现参数辨识,之后通过级联模块实现辨识结果的传递。该方案可同步辨识定子电阻、永磁体磁链和定子电感等3个电机参数,从而为NDO提供参数支持,减小因参数偏差引起的补偿误差。【结果】NDO能够实现高精度误差观测。此外,CMRAPI方案能够实现电阻、磁链和电感的准确辨识。【结论】仿真结果表明,电机长时间运行时NDO能够精确补偿由逆变器非线性特性引起的扰动电压,从而有效抑制输出电流畸变和转矩脉动。

Abstract: [Objective] The nonlinear characteristic of voltage source inverter causes output current distortion, which in turn leads to motor output torque pulsation and affects the control accuracy and stability of permanent magnet synchronous motor drive system. To solve this problem, an online compensation strategy based on the nonlinear disturbance observer (NDO) is proposed at the paper. [Methods] Firstly, the nonlinear factors such as inverter dead time and switching tube voltage drop were quantitatively analyzed. Then the NDO was constructed based on the synchronous rotating coordinate system current equation, the saturation gain function was designed, and the dq axis reference voltages were corrected by the feed-forward compensation channel. However, the NDO was highly sensitive to parameters. Thus, a cascade model reference adaptive parameter identification (CMRAPI) scheme was proposed. And the cascade module parameter identification task was decomposed into two groups of mutually independent identification tasks of stator resistance and permanent magnet magnetic chain as well as stator inductance, and each group of tasks realizes parameter identification through the model-referenced adaptive structure, and then realizes the transmission of the identification results through the cascade module. The proposed scheme could simultaneously identify three motor parameters such as stator resistance, permanent magnet chain and stator inductance, so as to provide parameter support for the NDO , and to reduce the compensation error caused by parameter deviation. [Results] The NDO was capable of high-precision error observation. In addition, the CMRAPI scheme can achieve accurate identification of resistance, magnetic chain and inductance. [Conclusion] The simulation results show that the NDO is able to accurately compensate the disturbance voltage caused by the nonlinear characteristics of the inverter during the long-time operation of the motor, thus effectively suppressing the output current distortion and torque pulsation.

黑龙江省自然科学基金(LH2024E092)