|
| 摘要: |
| 综述了发电机励磁系统参数辨识的方法及其在国内外的应用和发展,将辨识方法分为时域法、频域法和人工智能法3种。时域法原理简单,计算方便,对激励信号没有特别要求,容易实现,但是该方法没有滤波功能;频域法利用快速傅里叶变换,将时域上的卷积转化为频域上的简单乘积,计算方便,但它需要伪随机信号作为激励信号,对伪随机码的参数选取要视具体情况而定,对低阶系统的参数辨识准确度高;人工智能方法原理简单,对激励信号没有特殊要求,能辨识非线性系统,可以直接得到实际参数,但它也没有滤波功能,而且对系统的先验知识要求较高,这些先验知识制约着用遗传算法辨识系统参数的精度。3种辨识方法没有绝对的最优,可根据不同情况选用不同方法,必要时3种方法可结合使用。 |
| 关键词: 励磁系统,参数辨识,时域法,频域法,人工智能法 |
| DOI： |
| 分类号:TM761.11 |
| 基金项目: |
|
| Overview of parameter identification methods for generator excitation systems |
|
SHU Hui WEN Jinyu
|
| Abstract: |
| Parameter identification methods of generator excitation system and their applications and developments at home and abroad are overviewed.They are classified as time domain method,frequency domain method and artificial intelligence method.The time domain method is of simple principle,convenient calculation and easy implementation,without special request for test pumping signal and filter function.The frequency domain method transfers convolution calculation in time domain to product calculation in frequency domain by fast Fourier transform,making the calculation easier,but it needs pseudo-random code as the test pumping signal,whose parameters can not be specified easily.It has higher accuracy for lower order system.The artificial intelligence method is of simple principle,without special request for test pumping signal and filter function.It can identify nonlinear systems and get system parameters directly,but it needs a lot of experienced knowledge,which limits its accuracy.None of the three methods is absolutely optimal,and one or more should be selected according to the given situations. |
| Key words: excitation system,parameter identification,time domain method,frequency domain method,artificial intelligence method |
