A GPU-Based Lagrange Multiplier Optimization for Dynamic Economic Dispatch Aiming to Reduce Wind Power Curtailment

Due to the large-scale integration and the high uncertainty of wind power, the curtailment of wind power happens from time to time. To some extent, the curtailment could be reduced by adjust short-term and ultra-short-term dispatching of power systems. In this paper, a mathematic model dynamic economic dispatch (DED) is set up aiming to reduce the curtailment of wind power in in power systems. Computation of the adjustment is needed to be completed in very short time because it will be carried out in following one more hour. Therefore, the technology of heterogeneous computing is introduced to accelerate the DED solving process of Lagrange Multiplier optimization, and graphic processor unit (GPU) is used to execute parallel computing multiple data with the same instruction threads. A case based on IEEE reliability system is used to test effectiveness and efficiency of the model and the algorithm. (paper)