交流电机单神经元模型参考自适应控制系统研究

【作者】 魏庆伟；

【导师】 于海生；

【作者基本信息】 青岛大学 ， 电力系统及其自动化， 2006， 硕士

【摘要】 以交流电机中的异步电机、永磁同步电机为控制对象,重点研究了具有自学习和自适应功能的单神经元模型参考自适应速度控制器的理论和仿真实现方法。 首先对两种交流电机的数学模型、神经网络基本知识、模型参考自适应控制、矢量控制基本原理和电流滞环脉宽调制方法作了较为详细的介绍,并给出了传统交流电机矢量控制系统和神经网络模型参考自适应控制系统的结构图。 针对传统矢量控制系统中PID控制器的三个参数难以调整、电机参数改变或者外部环境变化时需要不断的对参数进行修正而带来的不方便,以及神经网络模型参考自适应控制系统中神经网络控制器理论复杂、难以实现的缺点,提出了用单神经元模型参考自适应控制器来代替传统PID控制器,利用单神经元的自学习、自适应能力,在线实时调整单神经元参数,从而使系统具有较好的自适应性能。并根据交流电机具有大惯性环节的特点改进了模型参考控制中的误差函数,将速度导数考虑进误差函数,使电机速度跟踪更平稳、迅速。 在理论分析的基础上,最后采用仿真软件MATLAB分别建立了异步电机和永磁同步电机的单神经元模型参考自适应控制系统仿真模型。通过大量的仿真试验表明:单神经元模型参考自适应控制器具有结构简单、可调参数少(只有一个微分参数k_d)、易于工程实现以及控制输出平稳的特点,并具有参数及负载时变的自适应能力,具有广泛的应用前景。更多还原

【Abstract】 This thesis is focused on the theory and simulation implementation of a single neuron model reference adaptive control (MRAC) controller provided with the ability of self-learning and self-adaptive, with the induction motor and Permanent Magnet Synchronous Motor (PMSM) taken as a control plant.Firstly, the math model of induction motor and PMSM, basic knowledge of neural networks and MRAC, theory of field-oriented control and hysteresis-current controlled PWM are introduced particularly, and the structure of vector control and neural networks MRAC is also provided.In allusion to the difficulty of parameters adjustment in traditional PID controller and the dis-commodiousness of constant modify when motor’s parameters or environment changes, the neuron MRAC approach is proposed in place of traditional PID controller, making use of it’s self-learning ability. A new error function is presented to avoid the affection of big-inertia of AC machines and even more to ensure better motor speed tracking efficiency.Based on the theory, then the simulation system of induction motor and PMSM based on MATLAB is proposed. The simulation results demonstrate that the scheme can ensure the motor speed tracking the reference speed quickly and precisely, and reduce its sensitivity to load disturbances. Furthermore, the algorithm is also robust for induction motor parameters variation. The designed controller is so simple that it is very easy to be implemented and applied.更多还原