A hybrid method for the dynamic parameter identification of generators via on-line measurements

To maintain the reliability and security of power system, the Independent Power Producers (IPPs) are required to provide the accurate dynamic parameters of the generation facilities to the Independent System Operator (ISO) or Regional Transmission Organization (RTO). Dynamic parameter identification which aims at obtaining accurate dynamic parameters has been one of the central topics in power system studies for years. Sensitivity analysis is the most popular and traditional method in dynamic parameter identification of power system. However, its effectiveness is highly dependent on the preset initial guess. Some intelligent methods such as GA and ANN which can handle this problem usually require much more time and are complicated to be applied. This paper proposes a hybrid method combining particle swarm optimization and sensitivity analysis for dynamic parameter identification. The proposed hybrid method provides the right balance and trade-off between convergence and computation speed. Particle Swarm Optimization, a relatively new intelligent optimization method, is employed to find an approximate solution in the first step. Then the sensitivity analysis is run to achieve an accurate solution starting with the approximate solution obtained from PSO. This paper focuses on key parameters, pre-recognized by PSS/E simulation with historical data, to reduce the number of simulation cases. Also the parallel programming is used to take advantage of multiple core processors to significantly increase the computation speed. The simulation results show the validity and benefit of the hybrid method.