永磁同步电机具有控制特性好、功率密度高、效率高、功率因数高等优点，是数控机床、机器人及诸多航天、国防设备的核心执行部件。在这些典型应用领域，如何进一步提升控制的动态响应和精度是其中的难点问题和研究热点。作为电机控制的内环，电流控制又是解决上述难点问题的关键。为此，本文针对永磁同步电机电流内环的高性能控制方法开展研究，主要内容包括电流动态性能的提升、电机参数辨识、转子转速与电流的滤波方法、死区补偿等。基于对永磁同步电机电流环传统预测控制策略的参数敏感性分析，提出增强预测控制策略鲁棒性的改进方法，在一定程度上降低参数不准确带来的影响。为提高电流环动态性能，利用一般系统电流动态变化过程要远快于反电动势变化过程这一特征，在不同环节分别提出消除数字控制一拍滞后影响及提升电流响应速度的方法。在前向通道，推导出一种取代PI调节器的电流增量预测控制策略，并采用改进方法增强其鲁棒性。在反馈通道，推导出一种电流增量预测算法，用电流预测值取代检测值作为PI调节器的反馈输入，使调节器输出与反馈输入在时序上保持同步，提高电流动态性能。在输出环节，通过理论分析，推导出一种考虑一拍滞后延时影响的简易改进方法及其相应的PI调节器参数设计规则。针对所提电流控制方法中使用的电机电感这一关键参数，提出一种基于增量式模型参考自适应的电感与惯量在线辨识算法，通过简化数学模型，降低辨识算法的计算负荷以及对电机参数的依赖程度。针对辨识结果出现的定向漂移等问题，采用判断激励信号有效性以及增大离散时间间隔长度的改进策略，消除噪声及未建模动态等非理想因素的影响。提出在电流增量预测控制策略中采用电感的在线辨识结果作为电感估计值，提高控制策略对参数变化的自适应性。针对测量环节引入的随机噪声，提出一种基于增量式卡尔曼滤波器的永磁同步电机电流滤波算法，有效降低滤波算法的计算量及对电机参数的依赖程度，该方法同时适用于对电机转速反馈信号的处理。结合增量式卡尔曼滤波器与增量预测算法，可实现电流滤波的同时补偿死区引起的电压扰动，获得更准确的相电流波形。
Permanent magnet synchronous motor (PMSM) has several advantages, such as perfect characteristics, high power density, high efficiency, and high power factor. It is a core execution unit being employed in many applications, such as computer numerical control machine tool, robots, and many aerospace and defense devices, etc. In these typical applications, how to further improve dynamic performance and precision remains one of the difficult problems and research focus. As inner loop, excellent performance of current is critical to solve these problems. Current control methods with high performance of permanent magnet synchronous motor is studied in this dissertation, including control strategies to improve dynamic performance of current, motor parameter identification, filtering algorithms of speed and current, and dead-time compensation.Based on the parameter sensitivity analysis of traditional predictive current control strategy in PMSM, improved methods to enhance system robustness are proposed, reducing the impact of inaccurate parameters in a certain extent.In order to improve dynamic performance of current loop, one feature of general system is exploited that dynamic process of current is much faster than that of back electromotive force, and methods are proposed to eliminate the impact of delay caused by digital control at different stages of current loop, increasing current response speed. In forward stage, an incremental predictive current control strategy is derived to substitute PI regulator, and improved methods to enhance robustness are analyzed. In feedback stage, an incremental predictive algorithm of current is derived, in which the detection value of current is replaced by predicted value as the feedback input of PI regulator. Thus, the feedback keeps pace with output of the regulator, increasing the dynamic performance of current. At output link, based on theoretical analysis, a simple improved method considering the influence of one-step-delay in digital control is derived, as well as corresponding parameter design rules.To the inductance, which is a key parameter in the current control methods proposed previously, an online identification algorithm of inductance and inertia based on incremental model reference adaptive system (MRAS) is proposed. By simplifying mathematical model, computational burden and dependence on motor parameters of the proposed algorithm is reduced obviously. To the unexpected phenomena of the identification results, influence of non-ideal factors, such as noise and un-modeled dynamics, are eliminated by judging the validity of the excitation signal and increasing the length of the discrete interval. The online identification result of inductance is directly employed in the incremental predictive current control strategy to improve the system adaptability to parameter changes. For random noise introduced in measurement process, a filtering algorithm of PMSM current based on incremental Kalman filter is propose, which obviously decreases the computation burden and the dependence on motor parameters. This algorithm can also be applied to filter feedback signal of motor speed. Combining the proposed filtering algorithm with the incremental prediction algorithm of current, it is able to realize the aim of current filtering and dead-time compensation at the same time, achieving more accurate phase current waveform.
