Publication

Abstract : Load forecasting (LF) plays a crucial role in maintaining the efficiency, stability, and reliability of power systems. Accurate LF models integrate real-time load demand data and meteorological conditions, capturing load variations across weekly trends, daily cycles, and seasonal patterns. However, the high variability and unpredictability of meteorological factors like air temperature, relative humidity, wind speed, and precipitation levels make short-term electricity demand forecasting difficult. Advanced data-driven approaches like dynamic mode decomposition with control (DMDc) and Kalman Filter are used to deal with such uncertainties. This work compares DMDc and Kalman filters by emphasizing their ability to capture weather-related load variations. The MAE and RMSE tests show that the DMDc model performs much better than the Kalman filter. The DMDc model has an MAE of 19.47 and an RMSE of 30.53, while the Kalman filter has an MAE of 32.98 and an RMSE of 46.54. The proposed Dynamic Mode Decomposition with Control (DMDc) technique integrates meteorological factors as control parameters, enhancing its ability to identify temporal fluctuations in load demand more efficiently than the Kalman filter.