Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This paper introduces a spectrogram-based framework that transforms continuously recorded high-resolution waveform measurements into a temporal object detection problem, enabling the joint detection, localization, and classification of events in power system dynamics. By applying the short-time Fourier transform to create spectrogram images, the method captures transient and harmonic signatures more effectively than raw waveform data. Experimental results show that this approach significantly outperforms traditional detection methods based on raw time-series measurements, particularly in identifying single-phase disturbances and three-phase faults.
Spectrogram analysis can substantially enhance the accuracy of event detection in power systems, outperforming traditional methods by leveraging the unique features of transient signals.
Continuously recorded high-resolution waveform measurements provide rich information about fast power system dynamics. However, they require automated methods to identify events. This problem is addressed by developing a spectrogram-based framework to jointly detect, localize, and classify events in real-world continuously recorded waveforms at the terminal of an Inverter-Based Resource. We recast this problem as a temporal object detection problem on spectrogram images, as they capture the transient and harmonic signatures more explicitly than in raw waveform data. Each time-series waveform is transformed using the short-time Fourier transform, and the resulting per-channel spectrograms are stacked as a tensor for event detection. We benchmark this method against a detector operating directly on raw time-series measurements. Experiments on single-phase disturbances and three-phase faults demonstrate that the proposed spectrogram method consistently improves event detection, localization, and classification over the raw waveform baseline.