基于梯形模糊数的模糊最小一乘回归模型  

Fuzzy least absolute regression model with trapezoidal fuzzy numbers

作　　者：李俊红[1,2] 曾文艺[1] 

机构地区：[1]北京师范大学信息科学与技术学院,北京100875 [2]河北师范大学数学与信息科学学院,石家庄050024

出　　处：《系统工程理论与实践》2015年第6期1520-1527,共8页Systems Engineering-Theory & Practice

基　　金：国家自然科学基金(10971243);河北师范大学应用开发基金(L2015K01)

摘　　要：本文给出一种梯形模糊数的距离,并基于这种距离,建立了关于梯形模糊数的模糊最小一乘回归模型.针对输入、输出和回归系数分别为实数和梯形模糊数的三种情形,进行了详细的研究与分析.最后,通过3个实例验证了所提模型的有效性.数据模拟结果表明我们的方法具有较好的拟合效果.In this paper,a new kind of distance between two trapezoidal fuzzy numbers is introduced,and fuzzy least absolute regression model with trapezoidal fuzzy numbers is proposed based on this distance.Aimed at that inputs,outputs,and regression coefficients are real numbers and trapezoidal fuzzy numbers,respectively,fuzzy least absolute regression model is investigated in detail.Finally,three numerical examples are used to illustrate our model reasonable.The experimental simulation results show that our proposed model has good fitting effect.

关 键 词：梯形模糊数 最小一乘 回归分析 回归系数 

分 类 号：O212.4[理学—概率论与数理统计]