GPU Based Parallel Matrix Exponential Algorithm for Large Scale Power System Electromagnetic Transient Simulation

Jinli Zhao,Juntao Liu,Peng Li,Xiaopeng Fu,Guanyu Song,Chengshan Wang
DOI: https://doi.org/10.1109/isgt-asia.2016.7796370
2016-01-01
Abstract:In order to meet the demand of fast electromagnetic transient simulation for large-scale interconnected power systems, a new method of GPU based parallel matrix exponential algorithm for power system electromagnetic transient simulation is presented in this paper. Firstly, the hardware structure and programming model of GPU are introduced. Then the high data parallelism based on matrix exponential integration algorithm is proposed. Then, a simulation test is carried out for a wind farm system including 17 WTGs. The accuracy and efficiency of the proposed method are verified. The results show that the speed of parallel computing based on GPU is about 2 times faster than that of CPU and Matlab/SimPowerSystems.
What problem does this paper attempt to address?