The Parallel-in-space Simulation For Transient Stability In Power System

As the scale of the interconnection of power systems is enlarging increasingly, It has become an urgent issue to develop fast and efficient tool for on-line dynamic security analysis. It is obvious that the traditional serial computation could not satisfy the requirement of real-time simulation for the transient process of large power systems. The parallel-in-space computing for large power system transient stability on the Cluster system is studied in this dissertation. The main works are focused on three problems including the task partitioning, the parallel algorithm designing and the assessment for the speedup.(1) Based on factorization path tree partitioning, a new partitioning scheme for the simulation-in-space of transient stability is proposed considering the computing of the differential equation. This algorithm adopts the generator bus weight and a new performance index function for assessing the partitioning scheme in the original algorithm of factorization path tree partitioning to make the distribution of computing load between each processor more reasonable.(2) The parallel solving process for a large sparse linear equation which has the bordered block diagonal form is derived clearly in this dissertation, it has two parts including the parallel LU factorization and the parallel forward and backward substitutions. In addition, this paper also optimizes the computation process and the communication process in the parallel-in-space simulation of transient stability. Namely, the post-partitioning ordering process is adopted to decrease the computation of forward and backward substitutions because only the generator bus owns the nonzero injecting element in the network equation; the communication process in the parallel simulation is improved through the altering of the style of data transmission.(3) This paper advances a formula for estimating the speedup under the ideal circumstance and analyzes the factors that influence the efficiency of the parallel simulation. It also illustrates the key point that when the decreasing time consumed by the parallel part is equal to the increasing time consumed by the serial part and the communication part in the parallel simulation process, the speedup will reach the maximum. Meanwhile, this paper also evaluates the number of processors needed to achieve the maximum speedup when the 3k-bus power system is tested on the Cluster1350.The simulation results obtained from two different tests validate the advantages of...