Research On The Online Frequency Domain And Time Domain Identification Of Synchronous Generator Parameters

Synchronous generator is one of the core elements in the whole power system. The correctness of the model and the accuracy of parameters aren’t only affect the accuracy of power system analysis and calculation, but also affect the development of the grid control strategy. The online identification method combines online measurement with identification technology, and parameters are more correct which are obtained on the actual operating conditions. Therefore, a research on the online identification for synchronous generator parameters has an important significance.An online frequency domain identification method based on the impedance real part and imaginary part detached representation algorithm is presented herein. From the perspective of frequency domain, this paper proposes a Parker model-based synchronous generator impedance matrix transfer function model. The model is represented by measurable variables, including d-axis voltage and current, q-axis voltage and current, excitation voltage, excitation current and so on. An impedance real part and imaginary part detached representation identification algorithm is proposed to analyse the identifiability of basic parameters in the model. Then an adaptive sinusoidal digital filter algorithm is used to obtain frequency domain information of the signal. Besides, a particle swarm optimization(PSO) algorithm is used to obtain basic parameters of the model. And then practical parameters are obtained according to the relationship between practical parameters and basic parameters. Studies show that with the identification algorithm of impedance real part and imaginary part detached representation, the basic parameters of the model are uniquely identifiable when the steady-state equations are considered, avoiding the problem of multi-valued parameters and improving the identification precision. Tests on the MATLAB simulation platform verified that compared to another method, the model and parameter identification method used in this paper don’t rely on the experience value of parameters, and are of high precision.The Parker model-based identification method is improved, and an online time domain identification method based on the direct integration least square algorithm is presented herein. From the perspective of time domain, this paper proposes a Parker model-based synchronous generator state equation model, which is represented by measurable variables. It’s proved that basic parameters of the model are uniquely identifiable with an indirect method. Then an step identification strategy for synchronous generator basic parameters is put forward, and the direct integration least square algorithm is used to obtain basic parameters of the model. And then practical parameters are obtained according to the relationship between practical parameters and basic parameters. Studies show that with the direct integration least square algorithm, it avoids the problem of solving complex differential equations, and simplifies parameter identification method. And the identification efficiency is improved, too. Tests on the MATLAB simulation platform verified that compared to another method, the model and parameter identification method used in this paper are reasonable and effective. The method doesn’t need frequency signal injection,and only online disturbance sampling datas are needed for identification, which has lower requirements for equipments. Besides, the time-domain identification model and algorithm are used in a project example, and verified the correctness and validity of the model and method.