基于神经网络技术的异步电机控制

针对常用的传统PID控制在提高异步电动机调速性能方面存在的瓶颈问题,本文提出了一种新型的单神经元控制算法以实现异步电动机的矢量控制.在分析单神经元控制器基本工作原理的基础上,为了加快神经元权值的学习训练速度,采用梯度下降法与变步长法相结合的控制算法,并利用MATLAB软件建立新型单神经元控制器与异步电动机调速系统的仿真模型,在计算机上进行了仿真,最后在笼型异步电动机上进行了应用实验.仿真与实验表明,这种单神经元控制器具有良好的自学习与自适应能力,可以改善调速系统的动态与静态性能,提高异步电动机的跟踪能力.

Asynchronous Motor Drive Based on Neural Network Technology

As to the bottleneck question of improving the asynchronous motor control performance existing in traditional PID control used commonly,a new single neuron control algorithm is proposed to achieve asynchronous motor vector control in this paper. On basis of analyzing the basic operation principle of the single neuron controller, the control algorithm combining gradient descent method with changing step method is adopted in order to accelerate the study and train speed of right value of the single neuron,MATLAB software is utilized to set up the new neuron controller model and asynchronrus motor emulation model,emulation is carried out on computer ,application experiment is carried out on the asynchronous cage motor finally. Emulation and experiment indicate that this new kind of single neuron controller has the capability of selfstudy and self-adaptivity,can improve the mobile and static performance of the motor control system and can enhance the tracing ability of the asynchronous motor.