Identifying Homologous Transient Power Quality Events via Monitoring Data Mining

Monitoring power quality is essential for evaluating the performance of power systems. Transient power quality events (TPQEs), such as voltage sags and swells, represent significant challenges that can lead to substantial economic losses for both utilities and their customers. Despite the widespread installation of power quality monitors (PQMs) across various locations to capture the waveforms of TPQEs, it remains challenging to determine whether these TPQEs, recorded by different monitors, originate from the same source, particularly when the time synchronization of PQMs is inaccurate. Furthermore, to date, no method has been developed for identifying events from the same source, leading to redundancy and invalidation of data in subsequent analyses. In response, a method for identifying events from the same source is proposed, utilizing a framework that includes modules for clustering by occurrence time, mining mapping relationships, and evaluating waveform similarity. This framework is applicable to distributed networks, even those with unknown topology and parameters, requiring only previously recorded waveforms. It also features the ability to adaptively update its parameters for optimal identification based on historical data, even after network reconfigurations. The effectiveness and reliability of this method have been verified through the analysis of hundreds of field recordings.