欢迎访问《太阳能学报》官方网站,今天是 分享到:
ISSN 0254-0096 CN 11-2082/K

太阳能学报 ›› 2022, Vol. 43 ›› Issue (9): 226-235.DOI: 10.19912/j.0254-0096.tynxb.2021-0240

• • 上一篇    下一篇

基于神经网络的最优带宽风电并网自抗扰控制

周雪松1,2, 杨子明1,2, 马幼捷1,2   

  1. 1.天津理工大学电气电子工程学院,天津 300384;
    2.天津理工大学天津市复杂系统控制理论与应用重点实验室,天津 300384
  • 收稿日期:2021-03-07 出版日期:2022-09-28 发布日期:2023-03-28
  • 通讯作者: 周雪松(1964—),男,博士、教授,主要从事新能源与智能电网方向的研究。zxs_myj@163.com
  • 基金资助:
    国家自然科学基金面上项目(51877152); 天津自然科学基金(18JCZDJC97300)

OPTIMAL BANDWIDTH ACTIVE DISTURBANCE REJECTION CONTROL FOR WIND TURBINE GRID-CONNECTION BASED ON NEURAL NETWORK

Zhou Xuesong1,2, Yang Ziming1,2, Ma Youjie1,2   

  1. 1. School of Electrical and Electronic Engineering, Tianjin University of Technology, Tianjin 300384, China;
    2. Tianjin Key Laboratory of Control Theory and Application for Complex Systems, Tianjin University of Technology, Tianjin 300384, China
  • Received:2021-03-07 Online:2022-09-28 Published:2023-03-28

摘要: 为了提高风能转换系统在故障穿越期间直流母线电压的暂态特性,解决线性自抗扰控制(LADRC)中固定的带宽带来的观测器响应速度和系统抗干扰性之间的矛盾,在线性自抗扰控制(LADRC)的基础上提出一种基于神经网络的最优带宽线性自抗扰控制(LADRC-OB)策略。首先分析带宽对系统性能的影响,然后根据系统已知模型设计LADRC控制器,并利用BP神经网络算法通过直流母线电压的参考值与实际值之间的偏差调整网络的输出。而神经网络的输出为LADRC的2个重要参数——观测器带宽ω0和控制器带宽ωc,这可解决LADRC的参数整定问题。最后,将LADRC-OB应用于1.5 MW的风能转换系统仿真模型中,并与采用双闭环PI时的控制效果进行对比,验证LADRC-OB具有更好的控制特性。此外,还对LADRC-OB的稳定性进行分析。

关键词: 风能, 神经网络, 变流器, 线性自抗扰控制, 最优带宽

Abstract: In order to improve the transient characteristics of DC bus voltage in wind energy conversion system during fault ride through, solve the contradiction between the response speed of the observer and the anti-interference performance of the system caused by the fixed bandwidth in the linear active disturbance rejection control(LADRC), the optimal bandwidth linear active disturbance rejection control based on neural network(LADRC-OB) is proposed based on LADRC. Firstly, the influence of bandwidth on the system performance is analyzed. Then the LADRC is designed according to the known model of the system, and the output of the network is adjusted by using BP neural network algorithm through the error between the reference value and the actual value of the DC bus voltage. The output of neural network is two important parameters of LADRC, observer bandwidth ω0 and controller bandwidth ωc. This also solves the problem of LADRC parameter setting. Finally, the LADRC-OB is applied to the simulation model of the 1.5 MW wind energy conversion system, and compared with the control effect of double closed-loop PI, it is verified that LADRC-OB has better control characteristics. In addition, the stability of LADRC-OB is analyzed.

Key words: wind power, neural networks, power converters, liner active disturbance rejection control, optimal bandwidth

中图分类号: