Font Size: a A A

Research On Distributed Parallel Computing Algorithms For Power System Small Signal Stability

Posted on:2007-09-26Degree:DoctorType:Dissertation
Country:ChinaCandidate:F LiFull Text:PDF
GTID:1102360212966739Subject:Power system and its automation
Abstract/Summary:PDF Full Text Request
With the interconnection of power systems, inter-area low frequency oscillation occurs occasionally, insufficient damping of the weak-interconnected power grid has become a serious issue affecting the power system stability. On-line small signal stability assessment may prevent the system from the oscillation; however, it isn't included in Dynamic Security Assessment. It is mainly because the on-line data of the whole system can not be collected together and so far no algorithm is developed to deal with this situation.To deal with the geographical distributed on-line data and to achieve high performance, the strategy of the parallel algorithms proposed in this paper is to decompose the entire system into several subsystems and then the main research is as following:1. A parallel algorithm used to solve the linear equations, named 'Multi-port Inverse Matrix parallel algorithm' proposed by China EPRI, is illustrated in detail and applied to the eigenvalue algorithms to manage the distributed Jacobean matrix.2. The reason, procedure and what a role it plays in all the eigenvalue algorithms of the vector normalization are introduced detailedly, the most effective parallel implementation of the vector normalization is selected.3. Several convergent criterions are compared and the most effective and suitable way for parallel implementation is selected4. Three distributed parallel computing algorithms are proposed, i.e. the Inverse Iteration/Rayleigh Quotient Iteration algorithm, the Simultaneous Iteration algorithm and the Implicit Restart Arnoldi algorithm. To achieve the best parallel efficiency several different parallel implementations have been exhausted.5. Based on MPI, the algorithms proposed in this paper are developed in Fortran language on Linux platform and tested on PC clusters. Simulation results on a small scale system, a large inter-connected system and a 10000-bus system are shown.Based on such research, three small signal stability distributed parallel computing algorithms are developed and implemented, i.e. the Inverse Iteration/Rayleigh Quotient Iteration algorithm, the Simultaneous Iteration algorithm and the Implicit...
Keywords/Search Tags:Small Signal Stability Analysis, Distributed Computing, Parallel Algorithm, PC Clusters, Power System
PDF Full Text Request
Related items