Design and Analysis of a Self-Tuning Speed Controller for Permanent Magnet Synchronous Motors Based on the Neural Network
Design and Analysis of a Self-Tuning Speed Controller for Permanent Magnet Synchronous Motors Based on the Neural Network
复制标题
DOI：
10.1166/jctn.2015.3869
复制
发表时间：
2015-07
影响因子：
--
通讯作者：
Lisi Tian;Yang Liu;Jin Zhao
中科院分区：
文献类型：
--
作者：
Lisi Tian;Yang Liu;Jin Zhao
提出了一种新型的基于人工神经网络的永磁同步电机速度控制器。分析了永磁同步电机速度环的小信号模型，这是控制器结构的基础。利用梯度规则更新神经元输入的权值，并利用辅助再训练过程改善控制器的动态性能。与传统的智能控制算法相比，绕过了繁琐的线下训练。因此，不需要广泛的电机知识，大大降低了计算成本。为了验证神经网络速度控制器的有效性，进行了全面的仿真。与常规的PID控制器进行了比较，结果表明该控制器具有良好的性能和鲁棒性。文中还讨论了人工神经网络速度控制器在物理平台上实现的可行性。
A novel artificial neural network (ANN) based speed controller of permanent magnet synchronous motors (PMSM) is proposed in this paper. Small signal model of the speed loop of PMSM which is the foundation of the framework of the controller is analyzed. The gradient rule is utilized to update the weights of the neuron inputs, and auxiliary retrain process is used to improve the dynamic performance of the controller. Compared with traditional intelligent control algorithms, the cumbersome offline training is bypassed. Therefore, no extensive knowledge of the motor is required, and computation cost is reduced significantly. A comprehensive simulation was done to verify the validation of the ANN speed controller. Comparison with the conventional PID controller demonstrates the outstanding performance and robustness of the proposed controller. The feasibility of implementation of the ANN speed controller on physical platforms is also covered in this paper.