Analysis method on parameter identifiabilityfor excitation system model of generator

The parameter identification methods, which use the experimental data to identify the parameters of the excitation system model, are widely used in the power systems. Although the model parameters obtained by these methods can properly fit experimental data, the identification results of some parameters may be unstable. To address this problem, this paper proposes a conception called sub-frequency domain sensitivity, which can provide a reliable index to assess whether the model parameters are easy to be identified or not for a nonlinear system. Based on this conception, a new parameter identification algorithm is proposed. In this algorithm, the existence of relevant parameters is judged by establishing the time domain sensitivity array of parameters at first, and then the identified parameters are divided into two categories: well-conditioned and ill-conditioned parameters. Based on the original ill-parameter group, evaluation representatives of the parameters are readjusted according to the sub-frequency domain sensitivity of parameters, finally, a "divide and rule" strategy is used to identify parameters. Case study is undertaken based on the IEEE ST2A type excitation system. Analysis results reveal that the proposed method can improve the accuracy and stability of parameter identification results in comparison with the traditional identification method based on time domain sensitivity.