发电机励磁系统参数辨识的仿真研究

Simulation Study on Parameters Identification for Generator Excitation System

研究发电机励磁系统参数辩识问题,由于励磁系统是一个非线性系统,造成电力系统不稳定。传统时域或频域辩识方法不能辩识其非线性环节,导致励磁系统辩识的精度低。为了提高发电机励磁系统的辩识精度,提出一种神经网络的发电机励磁系统参数非线性辨识方法。以发电机励磁系统实际输入作为神经网络的输入,以实际励磁系统输出与神经网络输出之间的最小误差作为目标函数,通过不断调整神经网络的权值对神经网络模型进行优化,最后得到满足系统误差要求的发电机励磁系统参数。仿真结果表明,改进方法解决了传统辩识方法无法准确辩识励磁系统非线性环节的难题,有效提高了励磁系统的辨识精度。

Excitation system is a nonlinear system, and the traditional identification method can not solve the non- linear identification problem, so the identification accuracy is low. In order to improve the identification accuracy of generator excitation system, this paper presented a nonlinear generator excitation system parameters identification method based on neural network. The actual inputs of excitation system were taken as the inputs of neural network while the minimum error between the actual output and neural network output was taken as the objective function. The neural network model was optimized by continuously adjusting the weights, finally the optimal parameters were obtained. The simulation results show that the proposed method has solved the problem of the traditional identification method and improved the identification accuracy of excitation system.