针对目前同步电动机负载突变时,均采用强力磁的方式控制励磁电压,提出了一种基于神经网络的同步电动机过渡过程的控制,克服了以往采用强励磁方式使同步电动机强行拉回到同步运行状态而引起的系统超调过大的不利因素。在神经网络的训练过程中采用“理想加噪声”的训练方式,表明了神经网络具有良好的动态逼近能力和较高的可靠性;在负载变动比较大的情况下,给出了用神经网络控制同步电动机的过渡过程仿真曲线。仿真结果表明,神经网络励磁调节器对于运行状态或是系统参数变化阶段,均具有很好的自适应性,尤其对同步电动机突加负载时系统的超调和到达稳态的时间有了明显的改善。
The paper not only discusses the control based on neural network of synchronous motor' s dynamic transition phase when load suddenly changed, but also gives the simulation curves of control based on neural network of synchronous motor' s dynamic transition phase when load has a big sudden change. The dynamical approach ability and high reliability of neural network are given as well. The simulations show the effectiveness of excitation neural network controller for controlling nonlinear system and the high ability of adaptiveness and anti-disturbance.