随着国家重大专向“高档数控机床与基础制造技术”的实施，对永磁交流伺服系统的性能提出了愈来愈高的要求。本文首先通过对典型应用案例的分析，揭示了交流伺服系统对应用系统性能的影响，指出提高交流伺服系统性能的必要性。永磁交流伺服系统参数的正确整定对系统的性能影响极大，设计者和用户均难以实现这一目标。因此，如何实现参数的自整定是这一领域的研究热点问题。由于在多数应用系统中惯量会经常发生变化，且惯量变化对系统的动态性能影响很大，本文提出了一种转动惯量在线辨识方法，能够实现惯量变化的自动辨识。在此基础上，针对交流伺服控制器PI 参数自整定方法中，忽略了整定过程的周期性和所需的自学习能力，造成整定过程效率低、重复操作次数多、整定时间长问题,依据迭代学习控制，提出了一种交流伺服系统速度控制器PI 参数自整定控制方案，使控制器能够进行仿人操作，根据历史控制经验不断地改进学习增益。仿真和实验结果均验证了上述方法的有效性，且这些方法结构简单，易于实现，已在已研发的产业化项目中得到具体应用。用于伺服系统的永磁同步电机，一般具有较小的转动惯量，这有可能使得电机的机械时间常数要小于电磁时间常数。面对这样的机电时间常数的关系，电流环在动态过程中不能抑制速度环的扰动，电流的动态响应明显变差。对此本文加入一个反电动势补偿环节，以此来消除转速在动态过程中对电流环的影响。理论分析并结合仿真研究，对比有无补偿环节的性能，提高了动态响应，仿真结果验证了补偿措施的有效性。在永磁交流伺服系统中，电流环位于最内环，其环宽对整个系统（速度环/位置环）的环宽具有制约作用。因此，在不提高功率器件开关速度的前提下如何提高电流内环的宽度也是该领域的研究热点。本文深入分析了传统PWM 方式限制电流环宽度的原因，提出了新的改进型PWM 方式，在不提高开关频率的情况下，使电流环的带宽得到了显著提高，从而扩大了永磁交流伺服系统的频带，提高了动态响应，实验结果均验证了该方法的有效性。Abstract: With national significant expertise to the "high-quality CNC machine and basic manufacturing technology",more and more demands are made for performance of the permanent magnet AC servo system. In this paper, through the analysis of typical applications, reveals the ac servo system influence on the performance of the system and the necessity to improve servo system performance.Permanent magnet AC servo system parameters in correct setting have a great impact on system performance. It is difficult to achieve this goal for designers and users. Therefore, how to achieve parameters self-tuning is of hot issues of this research field. In most applications inertia often changes. And the changes of inertia have a great influence on the system dynamic performance. A novel on-line identification method of inertia is put forward, and it can achieve automaticidentification of inertia changes. At present, most methods of self-tuning of PI controller for AC Servo drive cause variousproblems, such as low efficiency, more iterative manipulation and more setting time, due to ignoring the periodic propertyand the necessary self-study ability of setting process. This paper presents a novel self-tuning of PI speed controller for ACservo drive based on iterative learning control. The proposed scheme make the controller improve learning gain according tohistorical experiences like human being. The feasibility and validity of the proposed scheme has been verified with simulation and experimental result and the arithmetic is concise and apt to realize. PMSM used in servo system, generally has relatively small inertia, which could make the electrical mechanical time constant less than electromagnetic time constant. Face with such relationship between mechanical and electrical time constant, the current loop in a dynamic process can not curb the disturbance of speed loop, and dynamic response of current maybe bad. In this paper, add a link of speed compensation in order to eliminate the impact of speed on the current loop in dynamic process. Theoretical analysis and combine with the simulation studying, contrast the performance with and without the compensation link, simulation results show the effectiveness of compensation measures.In permanent magnet AC servo system, the current loop located in the inner ring. The bandwidth of the entire system is restricted by current loop bandwidth. Therefore, how to expand the bandwidth of the current inner loop under not raising power device switching speed condition has been a hot research topic. Based on the analysis of current loop bandwidth restricted conditions using traditional PWM style, an improved version of PWM mode is proposed. Current loop bandwidth has been significantly improved without raising switching frequency. Thus increase the permanent magnet AC servo system bandwidth and improve the dynamic response. Experimental results verify the effect of the method.关键词：交流伺服系统；转动惯量在线辨识；参数自整定；迭代学习控制；动态响应Keywords: AC servo system; inertia on-line identification; self-tuning; iterative learning control; dynamic response