Fast voltage detection for a single-phase dynamic voltage restorer (DVR) using morphological low-pass filters

In this paper, we introduce a Mathematical Morphology(MM) theory based low-pass filter to preprocess the single-phase voltage sag signal, to filter out noises and any other interferences in the sampled signal. A comparison between the MM filter and traditional low-pass filters is carried out to analyze the applicability and accuracy of the MM filter. The selection of structural element is discussed as well according to the filtering result. In addition, we introduce a "Specific Harmonic Elimination" method to eliminate certain low frequency harmonic of the sampled voltage after dq transformation, which could greatly simplify the design of the low-pass filter. Then a elliptic filter is used for constant component abstraction. Software simulation result has verified the practicability of the algorithm.

I. Introduction

Many practices have shown that single-phase voltage sags are the most frequent and most serious transient power quality problems in distribution power system. Dynamic Voltage Restorer (DVR) is an effective solution for transient Power Quality(PQ) problems. When using DVR to fast compensate voltage sag, the characteristics of fundamental wave need to be detected quickly, including the beginning and ending time of voltage sag, the sag voltage amplitude and the accompanying phase angle shift. The realtime capability and accuracy of the detection finally decide the performance of the compensation. And the design of a low-pass filtering algorithm also plays a decisive role. Its types, order and the cut-off frequency have great impacts on the dynamic response and accuracy of the detection algorithm. But detection accuracy and dynamic response speed can not be reached simultaneously. The decrease of the dynamic response time means the reduce of detection accuracy, and vice versa.