基于神经网络和预测控制的感应电机转速控制系统研究

发布时间：2019-05-23 12:49

【摘要】：近年来,随着电力电子技术、控制技术和计算机技术的发展,用交流拖动控制系统取代直流拖动控制系统已成为不争的事实,其中感应电机在运动控制中得到了广泛应用。感应电机具有高阶次、耦合性和非线性等特点,矢量控制技术在电机控制系统中广泛应用,在此基础上寻找新的控制策略设计出具有高性能、强抗扰能力和强适应性的控制系统,是当今国内外研究的热点。近年来,模型预测控制技术具有多目标、多变量、强约束性等优势,能提高控制系统的跟踪性、稳定性和鲁棒性等特点,在运动控制中引起了广泛关注。在基本矢量控制伺服系统中速度控制器通常采用传统的控制器,应用和设计时要通过不断的调试控制器系数找到使系统达到最佳性能的参数,这就显得非常麻烦。而调节之后系数固定,当电机参数变化以及外部负载扰动等因素的影响下,原本的良好控制系统的就会大打折扣。这样人们开始寻求新的控制算法,智能控制技术具有很强的抗扰动性和自适应性,且参数能够自动调节,如神经网络、模糊控制等,把智能控制技术应用在来电机控制系统中意义非常重大。本文做了两个方面的研究。一方面根据模型预测控制技术的基本原理和特点,把模型预测控制的优点和按转子磁场定向的方法相结合,以三相电压型变换器和感应电机作为预测控制的研究对象,以静止αβ坐标系下的电流为状态变量进行寻优比较的基本思想,设计了电流预测感应电动机伺服系统,并对提高系统性能进行了较为深入的研究。另一方面针对外部负载扰动或转动惯量等因素对电机控制系统性能的影响,提出了神经网络自适应PI控制器,并把它应用速度环上。在Matlab/Simulink环境中搭建系统的仿真模型并进行验证,仿真结果证明了神经网络PI速度控制器和基本的控制器相比,当外界环境发生变化时具有很强的自适应性。通过仿真对比可知,在负载转矩和转动惯量发生变化时神经网络速度控制器能够使系统保持良好的控制性能。
[Abstract]:In recent years, with the development of power electronics technology, control technology and computer technology, it has become an indisputable fact to replace DC drive control system with AC drive control system, in which induction motor has been widely used in motion control. Induction motor has the characteristics of high order, coupling and nonlinear. Vector control technology is widely used in motor control system. On this basis, a new control strategy is found to have high performance. Strong disturbance rejection and strong adaptability control system is a hot research topic at home and abroad. In recent years, model predictive control technology has many advantages, such as multi-objective, multi-variable, strong constraint and so on, which can improve the tracking, stability and robustness of the control system, and has attracted extensive attention in motion control. In the basic vector control servo system, the speed controller usually adopts the traditional controller, and it is very troublesome to find the parameters to make the system achieve the best performance by constantly debugging the controller coefficients in the application and design. Under the influence of the change of motor parameters and external load disturbance, the original good control system will be greatly reduced when the coefficient is fixed after adjustment. In this way, people begin to look for new control algorithms. Intelligent control technology has strong immunity and adaptability, and the parameters can be adjusted automatically, such as neural network, fuzzy control and so on. It is of great significance to apply intelligent control technology to the control system of incoming motor. In this paper, two aspects of research have been done. On the one hand, according to the basic principle and characteristics of model predictive control technology, the advantages of model predictive control are combined with the method of rotor magnetic field orientation, and the three-phase voltage source converter and induction motor are taken as the research object of predictive control. Taking the current in static 伪尾 coordinate system as the state variable for optimization comparison, the servo system of current prediction induction motor is designed, and the improvement of system performance is deeply studied. On the other hand, aiming at the influence of external load disturbance or moment of inertia on the performance of motor control system, a neural network adaptive PI controller is proposed and applied to the speed loop. The simulation model of the system is built and verified in Matlab/Simulink environment. The simulation results show that the neural network PI speed controller has strong adaptability when the external environment changes compared with the basic controller. The simulation results show that the neural network speed controller can maintain good control performance when the load torque and moment of inertia change.
【学位授予单位】：哈尔滨工程大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TM346;TP273

【参考文献】