计算机应用 ›› 2016, Vol. 36 ›› Issue (5): 1336-1340.DOI: 10.11772/j.issn.1001-9081.2016.05.1336

• 人工智能 • 上一篇    下一篇

基于加权一个变量的一阶灰色模型的动态轨迹预测算法

江艺羡, 张岐山   

  1. 福州大学 经济与管理学院, 福州 350116
  • 收稿日期:2015-10-08 修回日期:2015-11-18 出版日期:2016-05-10 发布日期:2016-05-09
  • 通讯作者: 江艺羡
  • 作者简介:江艺羡(1983-),女,福建漳州人,博士研究生,主要研究方向:灰色系统、人工智能、数据挖掘;张岐山(1962-),男,黑龙江绥化人,教授,博士,主要研究方向:灰色系统、数据挖掘、管理信息系统、决策支持系统、物流管理、物流工程。
  • 基金资助:
    国家自然科学基金资助项目(61300104);福建省自然科学基金资助项目(2013J01230)。

Prediction algorithm of dynamic trajectory based on weighted grey model(1,1)

JIANG Yixian, ZHANG Qishan   

  1. School of Economics and Management, Fuzhou University, Fuzhou Fujian 350116, China
  • Received:2015-10-08 Revised:2015-11-18 Online:2016-05-10 Published:2016-05-09
  • Supported by:
    This work is partially supported by the National Natural Science Foundation of China (61300104), the Natural Science Foundation of Fujian Province (2013J01230).

摘要: 针对基于Kalman滤波的跟踪方法需要对噪声特性和轨迹的运动规律进行假设的不足,将新陈代谢一个变量的一阶灰色模型(GM(1,1))引入动态轨迹预测方法,提出一种基于加权灰色GM(1,1)模型的动态轨迹预测算法(TR_GM_PR算法)。首先,顺序截取预测点前不同长度的子轨迹,计算采用灰色GM(1,1)模型拟合各子轨迹的相对误差及相应的预测值;其次,对各子轨迹的相对拟合误差进行归一化处理,根据处理后的结果设置各子轨迹预测值权重;最后,将各子轨迹获得的预测值与其对应权重的线性组合作为轨迹未来运行趋势的最终预测结果。采用2000-2008年美国大西洋飓风数据进行实验,TR_GM_PR算法6 h的预测正确率为67.6056%,比基于模式匹配的飓风预测方法提高2.6056个百分点。实验结果表明, TR_GM_PR算法适用于轨迹短期预测。此外,该预测算法计算简单、实时性高,能够有效提高动态轨迹的预测正确率。

关键词: 一个变量的一阶灰色模型, 新陈代谢, 轨迹, 预测

Abstract: The noise assumption and motion assumption of trajectory should be demanded in dynamic trajectory prediction based on Kalman filter. In order to eliminate this insufficiency, the metabolism GM(1,1) model was introduced in dynamic trajectory prediction. Thus a prediction algorithm based on weighted grey GM(1,1) model (TR_GM_PR algorithm)was presented. Firstly, sub-trajectories with different length before forecasting point were cut out in order, then the relative fitting errors and predicted values of sub-trajectories were calculated using grey GM(1,1) model. Secondly, the normalization processing of relative fitting errors was carried out, and the weights of predicted values were set according to the result. Finally, using the linear combination of predicted values and their corresponding weights, the running tendency of trajectory in future was predicted. Experiments were conducted with the Atlantic weather Hurricane data from 2000 to 2008. Compared with hurricane trajectory prediction method with pattern matching, TR_GM_PR algorithm improves the prediction accuracy ratio of 6 hours by 2.6056 percentage points to 67.6056%. The experimental results show that TR_GM_PR algorithm is suitable for short-term trajectory prediction. In addition, the new algorithm has simple calculation and high real-time performance, and can effectively improve the prediction accuracy of dynamic trajectory.

Key words: Grey Model(1,1) (GM(1,1)), metabolism, trajectory, prediction

中图分类号: