Medium voltage dynamic voltage restorer with neural network controlled voltage disturbance detector

This paper describes the medium voltage high power DVR (Dynamic Voltage Restorer) with the new voltage disturbance detection method and the real-time digital voltage control. The new voltage disturbance detector was implemented by using the delta rule of the neural network control. Through the proposed method, we can instantaneously obtain the peak value of each phase voltage even under the single phase input voltage conditions. Compared to the conventional synchronous reference frame method, the proposed one shows the minimum time delay to determine the instance of the voltage sag or voltage swell event. Also real-time digital voltage control technique was adopted, where the inverter output filter capacitance voltage, the filter reactor current and the load current are sampled to calculate the inverter PWM command for the next sampling interval. By using digital control, the disturbance voltage can be compensated to the reference voltage level within two sampling intervals. The proposed disturbance detector and the voltage compensator were applied to the medium voltage high power DVR (22.9kV/4MVA) that was developed for the application of semiconductor manufacture plant. The performances of the proposed DVR control were verified through computer simulation and experimental results. Finally, conclusions are given.