电流型有源电力滤波器的神经网络滞环控制

Current Source Active Power Filter Control Using Neural Network Technologies

神经网络滞环控制就是将神经网络和滞环比较器这两种方法相结合的一种控制方法。滞环控制是控制方法中较为简单的一种,它不需要系统太多的参数,实时性较好,但用滞环控制方法时系统开关频率不固定,这不仅影响电子开关的使用寿命,而且降低了系统的稳定性。将二者结合的控制方法不仅可实现神经网络对快速变量的控制,而且可以提高滞环控制的性能。通过训练神经网络能够完成滞环比较器的功能,并可取代滞环比较器。给出了计算机仿真结果,神经网络通过DSP实现。

In this paper, the application of neural network to power converter control is discussed. A new hysteresis comparator constructed by using neural network is introduced. Hysteresis band control is an effective and simple control method. It can easily nm without many system parameters. But the switch frequency of system is not fixed. So it not only makes the system unstable but also may lessen the life span of the switches. The control method that combines the neural network technology with the hysteresis band technology has a high performance in response of current. Through training the neural network can learn the control rules by itself and can replace the real hysteresis comparator in power converter control. The computer simulation results are given in this paper and they can demonstrate the effectiveness of the proposed method. The neural network is realized by using DSP.