欢迎访问昆明冶金高等专科学校学报官方网站,今天是 分享到:

昆明冶金高等专科学校学报 ›› 2019, Vol. 35 ›› Issue (5): 37-44.DOI: 10.3969/j.issn.1009-0479.2019.05.006

• 资源开发与测绘 • 上一篇    下一篇

不同损失函数约束下的加权灰色模型在沉降监测中的应用

王建英   

  1. 云南经济管理学院工程学院,云南昆明650106
  • 收稿日期:2019-09-19 出版日期:2019-10-31 发布日期:2019-10-31
  • 作者简介:王建英(1979-),女,河南商丘人,副教授,工学硕士,主要从事GPS、工程测量及安全监测研究。

Application of Weighted Grey Model under Different Loss Function Constraints in Settlement Monitoring

WANG Jianying   

  1. Engineering School,Yunnan University of Business Management,Kunming 650106,China
  • Received:2019-09-19 Online:2019-10-31 Published:2019-10-31

摘要: 加权灰色模型的权值确定方法及权值大小直接影响监测拟合及预测的可靠性。以采空区沉降监测为研究对象,通过平均绝对误差(MAE)、平方和误差(SSE)、平均相对误差(MAPE) 和均方相对误差(MSPE)4种不同损失函数约束下的定权方式,体现各监测值对模型的贡献率,确定不同的加权模型。结合工程实例,验证了加权灰色模型的可行性和可靠性,也体现了模型的时效性。同时经过拟合和预测对比,其4种损失函数约束加权模式下的灰色模型都能预测短期数据,但从长期效果看,采用MSPE更优。

关键词: 加权灰色模型PGM (1, 1), 损失函数, 权值, 拟合, 预测

Abstract: The weighting method and weight value of the weighted gray model directly affect the reliability of monitoring fitting and prediction.In this paper,the settlement monitoring of goaf is taken as theresearch object,and the weights under the four different loss function constraints are mean absolute error(MAE),square sum error(SSE),mean average relative percent age error(MAPE) and mean square relative percentage error(MSPE).The method reflects the contribution rate of each monitoring value to the model and determines different weighting models.Combined with engineering examples,the feasibility and reliability of the weighted grey model are verified,and the time liness of the model is also reflected.At the same time,through the fitting and prediction comparison,the gray models under the four loss function constraint weighting models can predict short-term  data,but in terms of long-term and effect,MSPE is better.

Key words: weighted grey model-PGM(1, 1), loss function, weight value, fitting, prediction

中图分类号: