毕业生知识库

       随着工业4.0概念的提出和中国制造2025战略的推进，高性能的工业机器人作为制造领域的核心装备成为了我国制造业转型升级迫切需要的国之利器。交流伺服系统作为工业机器人核心关键部件之一，其性能优劣直接影响到工业机器人的装备水平。目前工业机器人用交流伺服系统主要被日本安川、日本松下、美国罗克韦尔和德国科比等欧美日系品牌垄断。因此，研究并开发具有自主知识产权的交流伺服系统，对我国工业机器人技术的进步和装备制造业的转型升级具有重要的理论意义和实用价值。 论文结合国家十二五重大专项（No.2013ZX04007-011）、“机器人关键技术研究及产业化应用”专项（No.161100210300）和“高精度控制系统研发”专项（No. 2012HZ00006-4），针对工业机器人对交流伺服系统的技术需求，围绕工业机器人交流伺服系统中参数在线辨识算法、电流环控制策略、速度环控制策略、位置环控制策略及高速工业实时总线接口等关键技术进行了深入研究，完成的主要工作如下： （1）针对传统辨识算法额外注入d、q轴电流扰动引起电机转矩和磁链波动的问题，本文充分利用d、q轴电压离散稳态方程，提出一种单稳态下多参数在线辨识算法。该算法不需额外添加d、q轴电流扰动，利用电机正常加速和稳速运行过程中的数据构造了一组电压满秩方程，克服了电机模型欠秩和微分项干扰的问题，能同时在线辨识出定子电阻、电感和永磁体磁链3个电气参数。本辨识算法的离散化实现采用限定记忆最小二乘法代替递推最小二乘法，解决了因数据饱和导致辨识精度下降的问题。所提出的辨识算法不需增加新的硬件成本，而且在软件上采用递推算法易于数字实现。仿真和实验结果表明，所提出的辨识算法辨识精度高，收敛速度快，并且在不同工况下都具有很高的辨识精度。在电气参数辨识的基础上，完成了电流环PI参数整定和电压前馈解耦控制，实现电流环动态解耦控制。实验结果表明电流环动态解耦控制完全消除d、q轴耦合项，电流环具有良好的动态性能和稳态精度。

（2）针对传统无差拍电流预测控制对电机电气参数依赖性大及稳态下电压方程微分项引起系统振荡的问题，本文结合预测控制良好的动态性能和PI调节器优良的稳态精度，提出了一种改进的无差拍电流预测控制算法，并通过理论和实验验证了所提算法在电感参数容许误差、稳态精度、动态调节性能等方面得到了明显提升。将该算法与参数辨识算法相结合，可以进一步提高电流环的鲁棒性能，保证电流环快速且无超调的响应性能。 （3）针对固定参数PI调节器“快速性”和“超调”之间的矛盾，本文提出了一种变参数PI调节器，并在系统稳定性分析的基础上，给出了变参数PI调节器参数设计规则。针对负载突变对转速闭环的影响，本文提出了一种基于改进扩张状态观测器的速度控制算法，将扩张状态观测器观测到的负载扰动值作为变参数PI调节器的前馈补偿量。选用改进的对数函数作为扩张状态观测器的非线性函数，简化了扩张状态观测器的参数整定问题，为工程实现提供了可行方案。仿真和实验结果证明了所提算法的有效性。 （4）针对比例-前馈调节器在阶跃位置指令下易产生超调的问题，本文提出了基于自适应微分跟踪器的比例-前馈位置调节器。该调节器首先根据位置调节器的输出饱和程度决定是否给位置指令安排过渡过程；然后针对传统微分跟踪器物理意义不明确的问题，增加二阶微分跟踪器速度限幅环节；最后采用最小二乘法拟合微分跟踪器滤波因子h和输入阶跃脉冲数s的一次函数关系。与比例-前馈调节器和传统微分跟踪调节器相比，所提出的改进位置调节器具有优良的动态性能和稳态精度。实验结果验证了系统能对大范围位置阶跃指令实现无超调的快速响应，且具有自适应调节能力。 （5）基于以上技术，针对工业机器人高精度多轴同步控制的要求，设计了基于EtherCAT总线的交流伺服系统，并通过实验证明了所设计的交流伺服系统具有良好的动态性能和稳态精度，能满足工业机器人对交流伺服系统的要求。

With the development of “Industry 4.0” and the promotion of the strategy “Made in China 2025”, the high performance industrial robot, which serves as the core equipment in manufacturing field, has become more and more significant for the manufacturing industry in our country. Alternating-current (AC) servo system is one of the key components of industrial robot, whose performance directly affects the level of industrial robot equipment. Currently the key technologies of AC servo system for industrial robot are monopolized by a few foreign companies such as Yaskawa, Panasonic, Rockwell and Kobe, etc. Hence developing our own AC servo system is of great theoretical and practical value for the progress of industrial robot technology and the improvement of equipment manufacturing industry. This thesis is supported by the national 12th Five-Year technology major project under Grant No.2013ZX04007-011, the key technology research and industrialization application under Grant No.161100210300 and the high-precision control system research under Grant No.2012HZ00006-4. In order to satisfy the technical requirements of AC servo system for industrial robot, this dissertation researches deeply on the key technical parameters of AC servo system for industrial robot, including online identification algorithm, current loop control strategy, speed loop control strategy, position loop control strategy and high speed real time industrial bus. The main contributions of this thesis are as follows:

(1) In order to solve the problem that the infusion of d and q axis current disturbance into permanent magnet synchronous motor (PMSM) may lead to the torque and the flux fluctuation ripple, the online identification based on the d axis and q axis voltage equations under steady state is proposed in this thesis. The full rank equation is constructed based on the data collected in the normal acceleration and steady-speed operation of the motor, which would avoid the problem of deficient rank of motor model and differential disturbance. It would obtain the stator resistance, the inductance and the flux at the same time. The discretization of the identification algorithm uses the recursive fixed memory least square method instead of the recursive least squares method to solve the problem that the accuracy of data may reduce with the saturation of data. It does not require the new hardware and is easy to implement in software using the recursive algorithm. Simulation and experimental results show that the proposed algorithm has high accuracy and good convergence under different operating conditions. Moreover, the PI parameters tuning of current loop and the voltage feed forward decoupling control are completed based on the identification algorithm. The experimental results show that d axis and q axis coupling can be completely eliminated and the current loop can achieve good dynamic performance and steady state accuracy based on dynamic decoupling control of current loop. (2) In order to solve the problem that the traditional predictive current control strategy based on dead-beat control depends greatly on the electrical parameters of motor and differential term of voltage equation under steady state may cause the system oscillation, an improved current predictive control algorithm based on dead-beat control is proposed in this thesis, which is combined with the advantages of predictive control and PI regulator. The experimental results show that the proposed algorithm improves the tolerance range of inductance parameters, steady state precision and dynamic regulation performance. It can further improve the robust performance of current loop and ensure the fast response without overshoot by combining the algorithm with the parameter identification algorithm. (3) The traditional PI regulator which utilizes fixed parameters cannot well guarantee the requirement for both response and overshoot at the same time. In order to solve this problem, the variable structure PI regulator is proposed in this thesis. Moreover, the parameter design rule of the variable structure PI regulator is given based on stability analysis of the control system. In order to solve the influence of load mutation on the speed loop, the speed control algorithm based on an improved extended state observer is proposed, which takes the observed load disturbance value as the feedforward compensation for the variable structure PI regulator. An improved logarithm function is used as the nonlinear function of the extended state observer, which simplifies the parameter tuning and makes the engineering realization become possible. The simulation and experimental results demonstrate the feasibility and advantage of the proposed algorithm.

(4) In order to solve the problem that the proportional-feedforward regulator may produce overshoot under the step position command, the proportional--feedfoward position control regulator based on an adaptive tracking-differentiator is proposed in this thesis. First, according to the saturation of the position regulator, it would decide whether to set a transitional period for the output of position commands. Then, to solve the problem that the physical meaning of traditional tracking-differentiator is uncertain, a speed limit item constructed as second-order tracking-differentiator is added. Finally, it uses the least square method to fit the linear function relation between pluse step input s and filtering factor h in transient process. Compared with the proportional-feedforward regulator and the traditional differential tracking regulator, the proposed improved position regulator has excellent dynamic performance and steady-state accuracy. The simulation and experimental results show that this regulator can adaptively choose the optimal parameters of the transient process according to the different curve of pluse step and it can achieve fast response without overshoot for the wide range of step commands. (5) Based on the above key technologies, the AC servo system based on EtherCAT is designed to satisfy the high-accuracy synchronous control requirement for the multi axis of industrial robot. The experiment results show that the AC servo system has good dynamic performance and steady precision, which can meet the requirements of the industrial robot. Keywords: AC servo system for industrial robot, Online parameter identification, Predictive control strategy based on dead-beat control, Extended state observer, Tracking-differentiator regulator.