Probabilistic Preassessment Method of Parameter Identification Accuracy with an Application to Identify the Drive Train Parameters of DFIG
Yuqing Jin,Chongjiang Lu,Ping Ju,Christian Rehtanz,Feng Wu,Xueping Pan
DOI: https://doi.org/10.1109/tpwrs.2019.2953666
IF: 7.326
2020-01-01
IEEE Transactions on Power Systems
Abstract:Parameter identification facilitates the improvement of the accuracy of power system simulation. However, for complex models of electrical equipment, it is impossible to simultaneously identify all the parameters. Inaccurate values of the nontarget parameters (NTPs) will affect the identification accuracy of the target parameters (TPs). However, no suitable methods are available in the parameter identification process to assess and handle this adverse impact. Therefore, a probabilistic preassessment method (PPM) is proposed to assess the possible identification accuracy of the TPs when the NTPs are inaccurate. PPM can provide more useful quantitative information than traditional sensitivity analysis for selecting the disturbance form, the observation variable, and the TPs that can be accurately identified. Then, a statistical identification process (SIP) is proposed to eliminate the dependence of the traditional parameter identification method on accurate NTPs. In SIP, parameter identification is repeated by using random values of the NTPs. Then, the mean value of the high-probability identification results is selected as the final result. Some of the identification results can be adjusted according to the PPM result to further improve the accuracy. The proposed methods were successfully used to identify the parameters of a two-mass drive train model of a DFIG wind turbine generator under the assumption that the generator and controller parameters are unknown.