Published in IEEE Access, 2025
This study presents a wearable IMU-based framework for analyzing and optimizing psychomotor learning by mapping motion trajectories to performance metrics, identifying benchmark clusters that define optimal performance, and guiding corrective actions, validated through a case study on the forehand shot in table tennis.
Published in 19th IEEE International Conference on Industrial and Information Systems (ICIIS 2025), 2025
This study proposes a deep learning approach using LSTM networks with attention mechanisms to reconstruct missing solar irradiance data from long-term meteorological records, enabling accurate long-term forecasting for photovoltaic (PV) systems. The model outperforms conventional LSTM methods, achieving high accuracy in case studies from Colombo and Hotevilla, thus supporting improved solar energy planning and investment decisions.
Published in 2024 Moratuwa International Conference on Electrical Engineering (EECon) (IEEE), 2024
This study reviews and compares machine learning and deep learning models, specifically using the PLAID dataset, for classifying appliance loads in Non-Intrusive Load Monitoring (NILM), analyzing how sampling frequency affects model performance to guide the selection of effective NILM techniques.
Published in 2024 Moratuwa Engineering Research Conference (MERCon) (IEEE), 2024
This study uses wearable IMU sensors to capture hand movements during welding, applying machine learning and deep learning to predict welding quality parameters (such as symmetry and path distribution) through regression analysis with minimal error.