A Method for Identifying Major Disturbance Sources in a Regional Grid

Considering power quality monitoring data characteristics, this paper proposes a method of identifying major disturbances impacting power quality based on data mining technology in a regional grid. First, power quality disturbance events are detected using dynamic thresholds. Meanwhile, support vector machines are employed to identify disturbance source type of each branch (usually, monitor points include branches that access disturbance sources and bus nodes in power quality monitoring system) which is not recorded in the grid account. Second, the data to be analyzed is extracted and preprocessed. Third, suspected related branches impacting the voltage distortion in problematic bus nodes are extracted with Pearson correlation calculation between voltage-related power quality indices in each problematic bus node and current-related power quality indices in each branch. The contribution of each suspected related branch to bus nodal voltage distortion is quantified through partial correlation analysis. Finally, the sources with frequent occurrences are extracted as major disturbance sources by the means of calculating the frequencies of high-contribution sources based on monthly monitoring data. The proposed method is validated through a case study conducted on a regional power grid which includes multiple types of disturbance sources, such as railway tractions, wind farms and photovoltaic stations. It is proved that this method can effectively identify major disturbance sources that cause nodal voltage distortion, which provides an effective guidance for disturbance source governance.