华南理工大学学报(自然科学版) ›› 2014, Vol. 42 ›› Issue (4): 19-25,58.doi: 10.3969/j.issn.1000-565X.2014.04.004

• 动力与电气工程 • 上一篇    下一篇

基于执行依赖启发式动态规划的紧急电压控制

冯小峰 刘明波   

  1. 华南理工大学 电力学院,广东 广州 510640
  • 收稿日期:2013-12-02 修回日期:2014-02-19 出版日期:2014-04-25 发布日期:2014-03-03
  • 通信作者: 刘明波(1964-),男,教授,博士生导师,主要从事电力系统优化与控制研究. E-mail:epmbliu@scut.edu.cn
  • 作者简介:冯小峰(1984-),男,博士生,主要从事电力系统优化、运行与控制研究.E-mail:ucihqtep@163.com
  • 基金资助:

    国家自然科学基金资助项目(51277078)

Emergency Voltage Control Based on Action- Dependent Heuristic Dynamic Programming

Feng Xiao- feng Liu Ming- bo   

  1. School of Electric Power,South China University of Technology,Guangzhou 510640,Guangdong,China
  • Received:2013-12-02 Revised:2014-02-19 Online:2014-04-25 Published:2014-03-03
  • Contact: 刘明波(1964-),男,教授,博士生导师,主要从事电力系统优化与控制研究. E-mail:epmbliu@scut.edu.cn
  • About author:冯小峰(1984-),男,博士生,主要从事电力系统优化、运行与控制研究.E-mail:ucihqtep@163.com
  • Supported by:

    国家自然科学基金资助项目(51277078)

摘要: 针对长期电压稳定场景,应用执行依赖启发式动态规划方法设计紧急电压控制器,协调各种不同类型的控制手段来维持系统的长期电压稳定.利用每个控制周期节点的电压偏差二次型构建效用函数,对未来控制时间的效用函数累加建立代价函数,建立长期电压稳定控制的动态规划模型,采用 BP 神经网络建立评价网络和执行网络,根据 Bell-man 最优化原理来训练评价网络,用于近似出代价函数,并由执行网络产生最优控制量,进而求解问题.最后,以新英格兰 10 机 39 节点系统为例, 建立基于 PSAT 的仿真算例,结果表明所提的控制方案可行且有效.

关键词: 长期电压稳定, 紧急电压控制, 执行依赖启发式动态规划

Abstract:

Aiming at long- term voltage stabilization scenarios,an emergency voltage controller based on the action-dependent heuristic dynamic programming is designed and is applied to the long- term voltage stabilization of powersystems by coordinating different control schemes.In the investigation,first,an utility function is presented byusing the quadric form of the deviation between bus voltage and its reference value in each control period,and acost- to- go function is constructed by accumulating the utility function in future design.Then,a dynamic program-ming model for long- term voltage stabilization is established.Moreover,two BP neural networks are employed to de-sign the critic network and the action network,respectively,and the Bellman equation is employed to train the cri-tic network which approximates the cost- to- go function and the action network which produces optimal control out-puts,thus obtaining the solution of the dynamic programming model.Finally,a time- domain simulation of NewEngland 10- machine 39- bus system is carried out using PSAT,and the results verify the feasibility and effectivenessof the proposed method.

Key words: long- term voltage stabilization, emergency voltage control, action- dependent heuristic dynamic pro-gramming