【作者】 卢达；

【导师】 赵光宙；

【作者基本信息】 浙江大学 ， 控制理论与控制工程， 2013， 博士

【摘要】 二十世纪八十年代以来,永磁同步电机以其体积小、损耗低、可靠性高、维护方便等优点,被广泛应用于交流调速系统中。不断多样化和复杂化的应用场合,对交流调速系统的控制策略提出了更高的要求。本文通过对永磁同步电机调速系统的控制策略的深入研究,力图从多方向促进系统在高性能要求场合的应用。本文的主要工作和成果有：1、介绍了研究背景和基础。首先对交流调速系统控制策略的发展历程进行了综述,着重介绍了模糊控制、自抗扰控制和模型预测控制在交流调速系统中的研究和应用现状。其次描述了永磁同步电机的数学模型和两种基本控制系统,并对比了这两种系统的控制效果。2、详细分析了零矢量在模糊直接转矩控制系统中的作用原理和特点,改进了模糊直接转矩控制的模糊规则库、转矩隶属度函数和电压隶属度函数,保证了零矢量在模糊直接转矩控制中可以充分发挥保持电磁转矩的功能,有效减小了转矩脉动。仿真和实验结果验证了本文提出的包含零矢量的永磁同步电机模糊直接控制算法的有效性。3、针对近似离散最速反馈控制函数(fhan函数)作为自抗扰控制系统非线性控制算法时参数难于整定的问题,首先分析了fhan函数的控制机理,并据此提出了改进的自抗扰控制系统结构。这种结构明确了fhan函数中包含的可调参数的物理意义,从而为按照实际要求确定参数值提供了理论依据。当存在时滞或需要更大的稳定裕度时,只需对其中一个参数微调即可。因此,采用本文改进的自抗扰控制系统结构,大大简化了参数整定过程。仿真和实验结果验证了改进算法的有效性。4、针对传统模型预测控制计算量大、难以在采样周期较短的永磁同步电机调速系统中应用的难题,本文提出了递推模型预测控制策略。与传统模型预测控制中迭代求解操作变量不同,递推模型预测控制首先采用迭代学习算法求取操作变量第一项,随后采用递推Levenberg Marquardt算法求出操作变量其它项。分别计算了传统模型预测控制和递推模型预测控制的时间复杂度,并分析了递推模型预测控制的收敛性,从而理论上说明了提出算法的有效性。仿真结果表明本文提出的递推模型预测控制的计算量比传统模型预测控制大为减小,基于递推模型预测控制的永磁同步电机调速系统具有优良的动静态性能。更多还原

【Abstract】 Permanent magnet synchronous motors (PMSMs) are widely used in speed control systems since1980s due to their excellent features such as small size, high efficiency, high reliability and easy maintenance. As the applications of PMSM speed control systems are more diverse and complex than before, control algorithms used by speed control systems need improving to meet higher performance requirement. Therefore, an in-depth study on control algorithms of PMSM speed control systems is completed in this paper, which is expected to promote and expand the high-performance applications of PMSM speed control systems. The main contents are listed as follows.1. The research background is provided. Firstly, the history of control algorithms of AC speed control system is presented. Among the algorithms, fuzzy logic control (FLC), active disturbance rejection control (ADRC) and model prective control (MPC) are highlighted and their latest research results are surveyed. Secondly, the PMSM model and two most-used PMSM control systems are introduced. After that, the performance of the two control systems is compared.2. The working principle and features of zero voltage space vectors in PMSM fuzzy direct torque control (DTC) systems are analysed, based on which the rule set, torque fuzzy sets and voltage fuzzy sets of fuzzy DTC are improved to give full play to hold torque feature of zero voltage space vectors. The effectiveness of proposed algorithm is verified by simulations and experiments.3. ADRC adopts discrete time-optimal control synthesis (fhan) function that includes three adjustable parameters as the nonlinear state feedback controller (NLSEF). To simplify parameters tuning procedure, the control mechanism of NLSEF in ADRC is analyzed based on the characteristics of fhan function, and then a modified ADRC structure is proposed. In this structure, the meaning of parameters is clear and so need not turning. When time delay exists or more stability margin is required, tune the value of only one parameter a little larger than its calculated value can get satisfactory results. The effectiveness of proposed algorithm is verified by simulations and experiments.4. An optimization problem is required to solve online in MPC, which has heavy computational burden. On the other hand, the sample interval of PMSM control system is too short for a MPC algorithm. To apply MPC in PMSM control system, a recursive MPC (RMPC) algorithm is proposed in this thesis. In RMPC, iterative learning control (ILC) is used to obtain the first term of manipulated variables, and then recursive Levenberg Marquardt algorithm (RLMA) is adopted to solve the online optimization problem. In this control algorithm, the first term can be used as an effective control input, and the other terms guarantee a positive effect of the input to future behaviors of the system. Furthermore, the convergency of RLMA is analysed. Then, to testify that RMPC has less computational burden, time complexity of MPC and RMPC are computed. Simulation results show the effectiveness of proposed method.更多还原