Neural-Network Based Vector Control of VSCHVDC Transmission Systems
The application of high-voltage dc (HVDC) using voltage-source converters (VSC) has surged recently in electric power transmission and distribution systems. An optimal vector control of a VSC-HVDC system which uses an artificial neural network to implement an approximate dynamic programming algorithm and is trained with Levenberg-Marquardt is introduced in this paper. The proposed neural network vector control algorithm is analyzed in comparison with standard vector control methods for various HVDC control requirements, including dc voltage, active and reactive power control, and ac system voltage support. Assessment of the resulting closed-loop control shows that the neural network vector control approach has superior performance and works efficiently within and beyond the constraints of the HVDC system, for instance, converter rated power and saturation of PWM modulation.
S. Li et al., "Neural-Network Based Vector Control of VSCHVDC Transmission Systems," Proceedings of the 2015 International Conference on Renewable Energy Research and Applications (2015, Palermo, Italy), Institute of Electrical and Electronics Engineers (IEEE), Nov 2015.
The definitive version is available at http://dx.doi.org/10.1109/ICRERA.2015.7418673
2015 International Conference on Renewable Energy Research and Applications (2015: Nov. 22-25, Palermo, Italy)
Electrical and Computer Engineering
Center for High Performance Computing Research
Keywords and Phrases
Adaptive Dynamic Programming; Levenberg-Marquardt; Neural Network; Renewable Energies; Voltage-Source Converter; VSC-HVDC Transmission and Distribution
Article - Conference proceedings
© 2015 Institute of Electrical and Electronics Engineers (IEEE), All rights reserved.