Research On Power System Super-Real-Time Transient Stability Simulation And Real-Time Decision-Making Emergency Control System

As power systems become larger and larger and the components in the system become more and more complex,power system transient stability simulation(TSS)becomes more and more time-consuming,which seriously affects the working efficiency of power system planners and dispatchers.In this dissertation,fast algorithms for transient stability analysis and control are studied.Serial TSSs are accelerated by improving the sparsity techniques.Parallel TSSs are accelerated by improving the block bordered diagonal form(BBDF)method.Based on these studies,a real-time decision-making emergency control system is developed with the support of Guodian Nanjing Automation Co.,Ltd.The main achievements of this dissertation are as follows.A new heuristic ordering algorithm named approximate minimum degree,minimum number of source predecessors(AMD-MNSP)is proposed.Based on the minimum degree,minimum number of predecessors(MD-MNP),the index of source predecessor is introduced and the limit of minimum degree is appropriately relaxed during node ordering.The proposed ordering algorithm enhances the efficiency of the sparse vector method by reducing the number of nodes in the factorization path set of source nodes while maintaining the sparsity of the factorized matrix.A multi-path sparse vector method is proposed.According to the characteristics of source nodes,the source nodes are divided into two types,namely,type A and type B.The two types of source nodes form path sets separately.The iterative solution of the path set of type A source nodes is performed as normal with the fast forward substitution and the fast backward substitution.Unnecessary computation can be avoided by performing the fast forward and backward substitutions of the path set of type B source nodes only once during the iterative solution process of network equations.An efficient implementation of the nested BBDF(NBBDF)method based on subnet-core mapping and mixed programming of MPI and OpenMP is suggested.The composition of the parallel overhead of the BBDF method is analyzed in depth.The influence of the structure of CPU,the thread scheduling pattern,and the parallel API on the parallel overhead is tested in detail.An efficient mapping is formed between network topology,parallel communication topology,and CPU structure.The parallel overhead is reduced because the method gives full play to the advantages of the parallel algorithm and computer hardware.In the 24886-node power system modeled in detail,a speedup ratio of 17 x is obtained,and a 10-second simulation only takes 0.778 seconds.A fully parallel NBBDF(FNBBDF)method is proposed,which combines the NBBDF method proposed in this dissertation with the fully parallel BBDF(FBBDF)method.Three implementation schemes of the FNBBDF method are compared.The FNBBDF method breaks the theoretical efficiency limitation of 50% of the classic BBDF method while maintaining low parallel overhead,which raises the speedup and efficiency of parallel computing.In the 24886-node power system with detailed models,the speedup ratio is further improved to 21 x,and a 10-second simulation only takes 0.639 seconds.A detailed TSS-based real-time decision-making emergency control system is developed.The improved sparsity techniques based super-real-time serial TSS is adopted.Based on the trajectory characteristics of the stability restoration process of EPGs,the stability restoration criteria are proposed to significantly shorten the simulation time required for stability judgment.The emergency control system has passed the software testing of Jiangsu Software Product Testing Center and the product appraisal of Jiangsu Economic and Information Technology Commission.This work is an exploration of the real-time decision-making emergency control system and has a certain significance in the future research and application of real-time decision-making emergency control systems.