基于增量式模型的PMSM鲁棒电流预测控制技术

发布时间：2018-04-10 23:18

本文选题：永磁同步电机 + 鲁棒电流预测控制　；参考：《哈尔滨工业大学》2017年硕士论文

【摘要】：电流预测控制凭借其良好的动态响应性能成为电机控制领域近年来研究热点之一,特别是带有PWM调制模块的电流预测控制(本文中简称为PPC)。PPC是一种模型预测控制策略,控制性能依赖于电机模型参数的准确性,尤其是定子电感和转子磁链。当PPC控制器中电感和实际电感失配时,会造成电流的震荡,偏差较大时甚至会导致电流环的不稳定。当PPC控制器中磁链和实际转子磁链失配时,会造成电流静差,使控制系统过流或削弱系统的带负载能力。因此,在保证电流预测控制优点的同时,提高磁链和电感的参数鲁棒性是电流预测控制的首要挑战。目前的鲁棒电流预测控制算法通常比较复杂,需要包含多种观测器。此外,将算法移植到不同实验平台时,由于电机本体参数的不同,需要对多个参数进行调试才能获得满意的性能。针对以上问题,本文提出了一种无需转子磁链参数的鲁棒电流预测控制策略,称为R-PPC。首先,在永磁同步电机数学模型的基础上,推导出增量式数学模型,基于增量式模型设计了无需磁链参数的R-PPC控制器。并对R-PPC的权重矩阵和电流指令的选择进行分析。此外,还对R-PPC的控制器稳定性和电流静差进行了分析,从理论上证明了本文提出算法的有效性和可行性。接下来,在无需磁链参数的基础上,本文引入扩展状态观测器ESO以提高电感参数鲁棒性。根据带有电感误差的电机数学模型,建立了连续和离散的ESO。本文还对ESO进行了稳定性分析,并给出了观测器增益的选择方法。仿真结果验证了ESO可以准确的观测由于电感误差造成的电压误差,通过电压补偿,R-PPC可以有效提高电感参数鲁棒性。之后,在参数鲁棒性提高的基础上,为进一步提高电流性能,本文采用了自适应死区补偿策略。仿真和实验结果表明,该补偿方法与R-PPC结合可以实现良好的补偿效果。谐波含量大大减小,零电流钳位现象几乎消除。最后,对PPC和R-PPC进行了充分的对比仿真和实验验证,结果表明R-PPC可以无需磁链运行且对电感参数有很强的鲁棒性,突出了R-PPC相比于PPC的优势。此外,还在广州数控的伺服平台上进行了R-PPC的算法移植和实验验证,证明R-PPC算法的通用性和可移植性。
[Abstract]:Current predictive control has become one of the research hotspots in the field of motor control with its good dynamic response performance in recent years, especially the current predictive control with PWM modulation module (in this paper, PPC).PPC is a model predictive control strategy).The control performance depends on the accuracy of motor model parameters, especially stator inductance and rotor flux.When the inductance in the PPC controller mismatches with the actual inductor, it will cause the current oscillation, and even lead to the instability of the current loop when the deviation is large.When the flux in the PPC controller is mismatched with the actual rotor flux, the current static difference will be caused, which will make the control system overcurrent or weaken the load capacity of the system.Therefore, while ensuring the advantages of current predictive control, improving the robustness of flux and inductor parameters is the primary challenge of current predictive control.The current robust current predictive control algorithms are usually complex and require a variety of observers.In addition, when the algorithm is transplanted to different experimental platforms, because of the different parameters of motor body, it is necessary to debug many parameters in order to obtain satisfactory performance.To solve the above problems, a robust current predictive control strategy, called R-PPC, is proposed without rotor flux parameters.Firstly, based on the permanent magnet synchronous motor (PMSM) mathematical model, the incremental mathematical model is derived, and the R-PPC controller without flux parameters is designed based on the incremental model.The weight matrix and current instruction selection of R-PPC are analyzed.In addition, the stability and current statics of R-PPC controller are analyzed, and the validity and feasibility of the proposed algorithm are proved theoretically.Then, the extended state observer (ESO) is introduced to improve the robustness of inductance parameters without flux parameters.According to the mathematical model of motor with inductance error, continuous and discrete ESOs are established.The stability of ESO is also analyzed, and the method of observer gain selection is given.The simulation results show that ESO can accurately observe the voltage error caused by inductance error, and the robustness of inductor parameters can be improved effectively by voltage compensation.Then, based on the improvement of parameter robustness, an adaptive dead-time compensation strategy is proposed to further improve the current performance.Simulation and experimental results show that the compensation method combined with R-PPC can achieve a good compensation effect.The harmonic content is greatly reduced and the zero current clamping phenomenon is almost eliminated.Finally, PPC and R-PPC are compared and verified by experiments. The results show that R-PPC can run without flux linkage and has strong robustness to inductance parameters, which highlights the advantage of R-PPC compared with PPC.In addition, the R-PPC algorithm is transplanted and verified by experiments on the servo platform of Guangzhou NC, which proves the generality and portability of the R-PPC algorithm.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TM341

【参考文献】