基于改进多中心校正解耦内点法的动态最优潮流并行算法

基于改进的多中心校正(MCC)和解耦技术,提出一种求解动态最优潮流(DOPF)的并行算法。结合内点算法(IPM)框架与DOPF修正方程的分块箭形结构,给出修正方程的并行解耦-分解-回代解法。并结合这一解法特点,提出动态步长拉大技术及自适应最大校正次数技术,以单次迭代计算量小幅增加为代价,换取迭代步长的增大,迭代点中心性的提高,总迭代次数和计算时间的显著减少。解耦技术的使用,使得所提算法的核心计算都可并行完成。6～118节点系统的串行仿真结果表明,算法具有很好的鲁棒性和收敛速度,在多核集群系统上的并行仿真表明,算法具有理想的加速比和可扩放性,适合求解大规模的DOPF问题。

Parallel Algorithm of Dynamic Optimal Power Flow Based on Improved Multiple Centrality Corrections Decoupling Interior Point Method

This paper presents a new parallel algorithm to solve dynamic optimal power flow(DOPF) based on the methods of improved multiple centrality corrections(MCC) and decoupling. A parallel decoupling-factorization-substitution method for the correction equation of DOPF is proposed by integrating interior point method(IPM) framework and the block arrow correction equation,and then,the methods of dynamic increasing step length and adaptive corrections are given.A longer iteration step length and a better central point,which can be obtained by the proposed algorithm, give on the reduction in the number of iterations and savings in computing time than IPM.Most of operations in proposed method can be processed in parallel with decoupling.The sequential simulation results for systems that range in size from 6 to 118 buses show that the proposed method is fast and robust,and the simulations in multi-core cluster show that the method is very promising for large scale DOPF problems due to its excellent speed up and scalability.