Browsing by Author "Pelendagamage, S"
Now showing 1 - 2 of 2
- Results Per Page
- Sort Options
Item Embargo Estimation of Switching Overvoltages during Energization of Transmission Lines using Recurrent Neural Networks(979-833153728-9, 2025) Pelendagamage, S; De Silva, HOvervoltages frequently pose significant challenges during the energization of transmission lines. During restoration, transmission lines have to be energized from zero voltage to the nominal voltage and the switching of transmission lines is a primary source of these overvoltages. The magnitude and waveform of switching overvoltages are influenced by system parameters, network configuration, and the specific point in the wave cycle at which switching occurs. The ability to estimate peak overvoltages in real-time is crucial for operators, during power system restoration. Traditional methods rely on extensive simulations or empirical formulas, which may not provide the necessary speed or accuracy for operational decisions. It is crucial for operators to ensure that peak overvoltages from switching actions remain within safe limits. This paper introduces a compact long short term memory recurrent neural network (LSTM RNN) based methodology to estimate the peak overvoltages induced during line energization. The developed RNN is trained and tested using extensive simulated data in PSCAD. The results demonstrate that the proposed RNN technique accurately estimates the peak values of switching overvoltages, offering a reliable tool for operators during power system restoration.Publication Embargo SleepTune: A real-time adaptive wake-up support through EEG based neurophysiological monitoring for optimized sleep stage classification(Elsevier Ltd, 2027-03-30) Pelendagamage, S; Rathnayake, S; Munathanthiri, S; Edirisinghe, S.T; Jayakody, D. N. K; Munasinghe, P; Silva, B. N; Wijesinghe, R. E; Woo, Seong-TakSleep inertia (SI), cognitive impairment following awakening from deep sleep, affects daily performance. This study presents an EEG-based sleep stage classification system designed to avoid awakening from NREM3, the stage most strongly associated with severe sleep inertia, through neurophysiological monitoring. It employs a dual-electrode neoprene headband (F3–F4) with an OpenBCI Ganglion bio-amplifier streaming to an NVIDIA Jetson Nano running machine learning algorithms. Two complementary models were developed using the Human Sleep Project dataset (461,510 labeled epochs from 500 subjects), with 114 features extracted across time, frequency, and time–frequency domains from dual EEG channels: a Random Forest (RF) binary classifier for deep sleep detection and a Convolutional Neural Network (CNN) for overnight sleep profiling. The RF distinguishes NREM3 from an “Other” class (Wake, NREM1, NREM2, REM), achieving 94% accuracy (sensitivity 95% for NREM3, 93% for Other), while the CNN classifies all five stages with 87.8% accuracy (per-class sensitivity: Wake 92%, NREM1 71%, NREM2 92%, REM 89%, NREM3 95%). The smart alarm operates within a 30-minute pre-set wake-up window; the RF evaluates sleep stage every 30 s and triggers the alarm upon detecting a non-NREM3 stage (Wake, N1, N2, or REM), reverting to the set time with a 5-minute snooze if none is detected. Overnight wearable recordings demonstrated feasibility, with the alarm withheld during epochs classified as NREM3, and CUDA optimization reduced processing latency for embedded deployment. The framework supports shift workers, on-call professionals, and early-rising commuters; extends EEG-based sleep staging beyond specialized laboratories; and enables longitudinal sleep-architecture tracking to complement clinical assessment of irregular sleep patterns.
