Evaluation of Power Losses in Cable Caused by Harmonics

Power losses of cable induced by harmonics may lead to temperature rise, aging, insulation defect and even failure. Estimation of power losses based on Theoretical formula can not address the impact of cable condition uncertainly, thus, this paper proposed a method of harmonic-induced power losses estimation based on Backpropagation Neural Network with Levenberg-Marquardt algorithm. Historical measurement data of harmonic current and associated overall power losses are used to train the neural network. Power losses can be estimated under the new combinations of harmonic in different orders in the future. Case studies for low-frequency harmonics and supraharmonics are conducted. Through comparison with finite-element method, the results evidenced the effectiveness of proposed method.