导　　师： 皮佑国

学科专业： 081101

授予学位： 博士

作　　者： ;

机构地区： 华南理工大学

摘　　要： 主要研究永磁同步电动机的分数阶建模，研究围绕电动机分数阶模型存在性、电动机分数阶时域建模、电动机分数阶频域建模以及基于永磁同步电动机分数阶模型的分数阶比例积分控制器的速度控制研究四个方面展开。 永磁同步电动机分数阶模型的存在性是本文研究的首要问题。鉴于永磁同步电动机分数阶模型是否存在，目前尚无相关研究及其成果的报告，加之对分数阶微积分的物理机理和意义尚无统一的认识，因此很难从机理上去证明或者验证用分数阶微积分描述的电动机模型的存在。本文假设，如果分数阶模型存在，即分数阶模型比其整数阶模型更加贴近电动机的实际，那么采用依赖于模型的控制器对采用分数阶对象进行控制所得到的控制性能应该优于利用该控制器对整数阶对象进行控制的性能。本文结合仿真实验和原型实验研究的结果表明，永磁同步电动机的分数阶模型是存在的。 采用机理和数据相结合的建模方法对永磁同步电动机分数阶时域建模，提出电动机的一种分数阶模型，设计伪随机激励信号，获取实时实验数据并采用数值拟合方法来获取分数阶阶次，运用分数阶系统建模中Output-error辨识算法，实现数据建模。模型验证通过原型实验，由辨识得到的永磁同步电动机分数阶模型和传统的整数阶模型采用同一控制器设计准则，在相同的实验条件下比较系统的阶跃响应跟随性能。 采用机理和数据相结合的建模方法对永磁同步电动机分数阶频域建模，提出电动机的一种分数阶模型，由实时实验数据绘制出电动机的对数频率特性曲线。采用分数阶频域建模中经典Levy辨识算法，对算法在高频数据拟合好但在低频段拟合差的问题，用加权函数加以改进，得到永磁同步电动机分数阶模型辨识结果。模型验证通过原型实验，由辨识得到的永磁同步电动机分数阶模型和传统的整数阶模型采用同一控制器设计准则，在相同的实验条件下比较系统的阶跃响应跟随性能。 对永磁同步电动机时域和频域分数阶建模两种方法的过程进行了简述和比较，分数阶模型比整数阶模型对系统辨识频域数据拟合的更好。这或许可以解释电磁耦合的分布参数系统的特性不能用整数有限阶次建模，分数阶建模却可以得到更好的拟合结果。同时基于永磁同步电动机分数阶模型和整数阶模型，采用同一分数阶控制器设计准则，给出分数阶比例积分PIλ控制器的设计方法和过程，比较系统的阶跃响应跟随性能。 This paper presents the fractional order modeling for the permanent magnet synchronousmotor /(PMSM/). Study on the existence of fractional model for PMSM and fractional-ordermodeling for PMSM from time and frequency responses. The performance of step response isalso studied based on a fractional proportion integration controller which is designed by theproposed fractional model of PMSM. The existence of PMSM fractional model is primarily studied. Considering there is norelated research and report that whether the fractional-order for PMSM is exist or not, what ismore, there is no unified understanding of the physical interpretation for the fractionalcalculus, so it is difficult on the mechanism to validate the existence of fractional-order modelfor PMSM. This paper assumes that if the fractional order model of PMSM is exist, it ismeans fractional-order model is more precise to describe the PMSM than the integer model,in that way, using the same controller which is designed by the fractional-order to control thePMSM should have better performance than the integer model. Based on the results ofsimulations and real-time experiments, the existence of PMSM fractional model is validated. Time domain modeling of fractional-order system for PMSM is presented by adoptingthe combination of mechanism and data. A fractional model for the PMSM is proposed, thepseudo-random signals is designed to obtain real-time experiment data and numerical fittingto get the fractional order, then data modeling is realized by using the output-erroridentification algorithm of fractional-order system. In model validation, two proportionalintegral /(PI/) controllers are designed with the same scheme according tothe identifiedfractional order model and the traditional integer order one, and real-time experimental resultsare presented to demonstrate the advantage of the proposed fractional order model. Frequency domain modeling of fractional-order system for PMSM is presented byadopting the combination of mechanism and data.A fractional model for the PMSM isproposed, in order to identify the parameters of the proposed fractional order model, anenhancement of the classic Levy identification method with weights is applied. In a real-timePMSM velocity servo plant, the fractional order model is identified according to theexperimental tests using the presented algorithm. Two proportional integral /(PI/) controllersare designed for velocity servo using a simple scheme according to the identified fractionalorder model and the traditional integer order one, respectively. The experimental testperformance using these two designed PI controllers is compared to demonstrate theadvantage of the proposed fractional order model of the PMSM velocity system. Discuss and compare the results with two methods of fractional-order moding for PMSMin time domain and frequency domain. Experimental data is used for demonstration that thefractional order model fits much better for the system identification frequency data over theinteger order model. This may be explained by the nature of the distributed parameter systemof the electromagnetism coupling thus may not be captured by integer finite order modelingwhile a fractional order modeling can offer potential to perform a better fitting. With the samefractional controller design rules, two simple PIλcontrollers aredesigned using the phasemargin and gain crossover frequency specifications according to the identified fractional orderand integer order models, the performance of the step response is compared.

关 键 词： 分数阶 建模 系统辨识 算法 最小均方差法 分数阶比例积分控制器 永磁同步电动机

分 类 号： [TM341]

领　　域： [电气工程]