Emerging energy conversion systems are characterized by increased rates and magnitudes of transients due to distributed generation, feedback-controlled loads and new entities like microgrids. A characterization of power quality in transients is thus gaining in importance in all power networks. Dynamic phasors offer a natural way to extend metrics based on steady state quantities such as phasor magnitude and RMS values to transients. The widespread use of high-bandwidth sensors enables a characterization of both steady-state and transient operation. However, the volume of so generated data is such that it necessitates extensive pre-processing and extraction of events of interest in estimation and control. An important issue then becomes how to pre-process that input data set, hoping to avoid excessive storage, communication and computation requirements. At the same time, the goal is to retain key benefits of fast sampling, such as the fast detection of events, and use in future wide-area and model-based protection algorithms. In this dissertation we introduce a fast sparse alternative to the standard (FFT-based) evaluation of dynamic phasors, which bridges the gap between single-waveform-sample methods (such as Akagi's instantaneous power theory), and the standard approach, which requires a full-cycle of waveform data. Our "sub-cycle" approach utilizes a small number of waveform samples (as few as two samples) to evaluate a few dominant harmonics, in contrast to full-cycle processing, which evaluates all harmonics, regardless of their magnitude. The sub-cycle approach provides a flexible trade-off between computational cost, which increases with the number of time-domain samples used, and the attendant improvement in the resulting dynamic power quality metrics. We propose to use it to generate a dynamic decomposition of apparent power into physically-meaningful components, based on power quality metrics that are defined entirely in terms of dynamic sub-cycle phasors. This dynamic power decomposition captures transient behavior, and reduces to constant components in steady-state. Thus, the sub-cycle approach provides a platform that can be used to monitor transient and steady state behavior of faulted systems when detection speed is critical. It can be integrated with algorithmic (intelligent) monitoring systems to provide faster system-level protection. We expect a good performance in applications where the presence of harmonics serves the role of a fault signature, including diagnostics of components (drives and converters), forensics of energy grids, and disturbance source localization.
展开▼
