Predicting transmission losses using EEMD – SVR algorithm

Abstract

This work introduces a predictive model for evaluating transmission losses in the Java-Bali electrical system using ensemble empirical mode decomposition (EEMD) and support vector regression (SVR) techniques. Transmission losses, a critical aspect of energy efficiency, are affected by several operational aspects, such as load flow, energy composition, peak load, and meteorological factors such as transmission line temperature. Transmission losses data were decomposed into many intrinsic mode functions (IMFs) by EEMD, effectively capturing both high-frequency (short-term) and low-frequency (long-term) trends. The SVR algorithm, utilizing a radial basis function (RBF) kernel, was subsequently employed to predict the deconstructed IMFs, facilitating accurate predictions of transmission losses. The proposed EEMD-SVR model achieved a mean absolute error (MAE) of 5.43%, with the highest error observed during the period of abrupt load shifts. These results confirm the model’s strength in identifying long-term transmission loss patterns, making it suitable for system planning and operational forecasting. While the model exhibited high prediction accuracy, especially in recognizing long-term trends, it faced limitations in accurately predicting abrupt changes in transmission losses. Therefore, future improvements should aim to enhance responsiveness to sudden changes in the system dynamics. The result suggests that the EEMD SVR model can proficiently assist power system operators in monitoring and mitigating transmission losses.