华南理工大学学报(自然科学版) ›› 2009, Vol. 37 ›› Issue (11): 50-55.

• 动力与电气工程 • 上一篇    下一篇

电力系统可靠性原始参数的优化GM(1,1)预测

张勇军1,2  袁德富1,2   

  1. 1.华南理工大学 电力学院, 广东 广州 510640; 2.广东省绿色能源技术重点实验室, 广东 广州 510640
  • 收稿日期:2009-02-12 修回日期:2009-04-24 出版日期:2009-11-25 发布日期:2009-11-25
  • 通信作者: 张勇军(1973-),男,博士,副教授,主要从事电力系统优化规划与控制、可靠性和HVDC等研究 E-mail:zhangjun@scut.edu.cn.
  • 作者简介:张勇军(1973-),男,博士,副教授,主要从事电力系统优化规划与控制、可靠性和HVDC等研究.
  • 基金资助:

    国家自然科学基金重点资助项目(50337010);广东省自然科学基金资助项目(06025630)

Prediction of Original Reliability Parameters of Power System Using Optimized GM( 1,1 ) Model

Zhang Yong-jun 1.2   Yuan De-fu 1.2   

  1. 1. School of Electric Power, South China University of Technology, Guangzhou 510640, Guangdong, China; 2. Guangdong Key Laboratory of Clean Energy Technology, Guangzhou 510640, Guangdong, China
  • Received:2009-02-12 Revised:2009-04-24 Online:2009-11-25 Published:2009-11-25
  • Contact: 张勇军(1973-),男,博士,副教授,主要从事电力系统优化规划与控制、可靠性和HVDC等研究 E-mail:zhangjun@scut.edu.cn.
  • About author:张勇军(1973-),男,博士,副教授,主要从事电力系统优化规划与控制、可靠性和HVDC等研究.
  • Supported by:

    国家自然科学基金重点资助项目(50337010);广东省自然科学基金资助项目(06025630)

摘要: 考虑到可靠性原始参数的缺乏对电力系统可靠性评估结果的真实性和有效性影响很大,用优化的GM(1,1)模型预测可靠性原始参数,开发小样本系统.优化的GM(1,1)模型在以最小二乘法优化初值的基础上,分别求取不同时间段的原始参数序列的拟合数列,再以各拟合数列与原始数列之间的模糊贴近度为权重系数对预测值进行优化加权组合.此模型既能体现数据的最新变化态势,又能体现总体发展趋势,充分挖掘原始参数包含的信息量,克服传统GM(1,1)模型预测可靠性参数随预测点推移预测精度下降较快的缺点,尤其适用于新投入元件可靠性原始参数的多点预测.

关键词: 电力系统, 可靠性原始参数, 模糊贴近度, 优化GM(1, 1)预测

Abstract:

As insufficient original reliability parameters greatly reduce the authenticity and effectiveness of the reliability assessment results of power systems, an optimized GM ( 1,1 ) model is used to predict original reliability parameters for the purpose of exploiting some small sample systems. In the optimized GM ( 1,1 ) model, the fitting series of the original parameter series within different time is obtained based on the optimization of the initial value using the least square method, and the fuzzy nearnesses between the original series and the fitting series are used as weight coefficients to treat the prediction values with weighted optimization. This model reflects not only the latest changes but also the overall development trend of data. Therefore, it fully exploits the information contained in original parameters and solves the problem existing the conventional GM ( 1,1 ) model, namely the rapid decline of prediction precision with the prediction point. The proposed model is applicable to the multi-point prediction of original reliability parameters of new electricity components.

Key words: power system, original reliability parameter, fuzzy nearness, optimized GM ( 1,1 ) prediction