基于矢量量化的长期直觉模糊时间序列预测  

Long-term intuitionistic fuzzy time series forecasting based on vector quantization

作　　者：郑寇全[1,2] 雷英杰[1] 王睿[1] 余晓东[1] 

机构地区：[1]空军工程大学防空反导学院 [2]中国人民解放军68331部队

出　　处：《吉林大学学报（工学版）》2014年第3期795-800,共6页Journal of Jilin University:Engineering and Technology Edition

基　　金：国家自然科学基金项目(61272011;61309022);国家重点实验室开放基金项目(2012ADL-DW0301)

摘　　要：通过引入滑动窗口机制和矢量量化技术,较好地解决了直觉模糊规则零匹配的问题,准确反映了不确定时序系统数据的分布特性,提高了复杂环境下时间序列长期趋势预测的精度,扩展了直觉模糊时间序列预测理论的应用范围。最后通过典型实例验证了该方法的有效性和优越性。A new method for long-term Intuitionistic Fuzzy Time Series （IFTS） forecasting is proposed. By employing the sliding window scheme and vector quantization technique, the non- matching problem of intuitionistic fuzzy rules is solved efficiently. Meanwhile, the distribution characters of the uncertain time series data system are reflected accurately. The long-term forecasting accuracy of time series in the complex environment is improved, thus greatly extending the IFTS forecasting application. Finally, the experimental results validate the efficiency and advantage of the proposed algorithm.

关 键 词：计算机应用 直觉模糊集 时间序列 确定性转换 矢量量化 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]