Research Of The Switched Capacitor Active Filter And Hysteresis Control Based On The Neural Network

As the rapid development of power electronics ,all kinds of nonlinear load have been applied broadly in many fields. However it arouse the serious harmonic pollution in the power system. Because the harmonic affects the electricity quality of the power system, the way to eliminate the harmonic is considered more and more by scientists. Active power filters(APF) are utilized widely in the power system to compensate the harmonic. The main circuit of traditional APF includes large reservoir capacitors or inductors in DC sides in order to generate the constant voltage or current. Thus the traditional APF can generate the right compensation signal. But the large reservoir capacitors and inductors are not only expensive but also physically very large. What's more, the energy required to maintain this constant DC level is obtained from the supply via the inverter, and this limits the performance of the filter. This paper researched the switched capacitor active filter to overcome the disadvantages of the traditional APF. The structure of the switched capacitor filters is very simple. It is just the combination of switches and capacitors and inductors. The switched capacitor configuration removes the requirement for a large current or voltage source, which leads not only to reduction in cost but also in physical size. The switching frequency of the filter is also reduced.This paper also researched a new kind of control method. A feedforward neural network works as hysteresis comparator in the PWM control. The combination of neural network and hysteresis improves the performance of control. In order to verify the validity of the structure and control method which are researched in this paper, we apply MATLAB software to emulate the system. On this basis, we programmed on the TMS320F240 chip to realize neural network hysteresis control. The emulation and the experiment results verify the validity of the proposed scheme.