Fast parallelized algorithms for on-line extended-term dynamic cascading analysis
Fast parallelized algorithms for on-line extended-term dynamic cascading analysis
复制标题
用于在线长期动态级联分析的快速并行算法
DOI：
10.1109/psce.2009.4840238
复制
发表时间：
2009
期刊：
影响因子：
--
通讯作者：
J. McCalley
中科院分区：
文献类型：
--
作者：
S. Khaitan;Chuan Fu;J. McCalley
高速扩展项时域仿真（TDS）具有快速在线计算能力，能够预测系统对扰动的中期动态响应并识别纠正措施。针对计算速度的发展，本文提出了一种拟部署在超级计算机Blue Gene/L上的并行策略来模拟电力系统动力学，该策略可以被描述为一组微分代数方程（DAEs）。为了处理DAE刚度问题，充分利用显式和隐式积分方法的优点，采用了一种称为递归投影法（RPM）的划分算法。此外，采用波形松弛法（WRM）分离DAE的刚性部分，实现了并行计算的良好负载平衡。利用多正面大规模并行稀疏直接求解器（MUMPS）求解隐式方法中涉及的线性系统。本文报道了该系统的设计
Very fast on-line computational capability to predict mid-term dynamic system response to disturbances and identify corrective actions is an important attribute of high-speed extended term (HSET) time domain simulation (TDS). Focusing on the development of computational speed, this paper propose a parallel strategy intended for deployment on the super computer, Blue Gene/L, to simulate the power system dynamics, which can be described as a set of differential algebraic equations (DAEs). To deal with DAE stiffness problems and fully capture benefits of explicit and implicit integration methods, the partition algorithm called recursive projection method (RPM) is employed. Additionally, good load balancing for parallel computation is achieved using waveform relaxation method (WRM) to separate the stiff parts of DAE. Multi-frontal Massively Parallel sparse direct Solver (MUMPS) is utilized to solve the linear systems involved in the implicit methods. This paper reports on the design