控制理论(社会学)
定子
离散化
瞬态响应
谐波
瞬态(计算机编程)
职位(财务)
转子(电动)
计算机科学
估计理论
永磁同步电动机
谐波分析
反电动势
物理
稳健性(进化)
带宽(计算)
计算
电压
电感
观察员(物理)
稳态(化学)
同步电动机
噪音(视频)
卡尔曼滤波器
磁铁
频率响应
感应电动机
扭矩
国家观察员
梯度下降
作者
Jinxiang Deng,Hong Wang,Yingkai Xiao,Jiaxin Deng,Zhongxue Gan,Danyang Bao
标识
DOI:10.1109/jestpe.2026.3672124
摘要
In high-speed permanent magnet synchronous motor (PMSM) drive systems, low switching frequencies increase discretization errors in conventional sliding mode observers (SMOs). This results in greater harmonics in the estimated stator currents and back electromotive force (EMF), thereby degrading the accuracy of speed and position estimation. This paper investigates steady-state position errors of conventional SMOs under low carrier ratios. The analysis reveals that the primary causes are model discretization errors and quasi-sliding mode motion errors. To address these issues, exact discrete voltage equations are derived in the two-phase stationary reference frame. Subsequently, an improved discrete-domain SMO is developed, incorporating a redesigned back-EMF computation method to enhance speed and position estimation accuracy. Furthermore, to improve dynamic performance under high-speed operating conditions, an adaptive quadrature phase-locked loop (AQPLL) is introduced. This method employs a stochastic gradient descent (SGD) algorithm for online parameter tuning, expanding the bandwidth for faster transient response while maintaining noise immunity at low speeds. The proposed method is validated through experiments on a low-inductance PMSM, demonstrating high estimation accuracy, reduced transient errors, and robust performance under parameter mismatches at a carrier ratio as low as 7.
科研通智能强力驱动
Strongly Powered by AbleSci AI