一种新的基于GM(1,1)模型的粗大误差判别模型

New distinguishing model for gross error based on GM(1,1) model

针对未知概率分布的小样本数据识别问题,提出了一种新的基于GM(1,1)模型的粗大误差外推判别模型。介绍了判别原理和步骤,基于GM(1,1)建模精度和检测值精度确定判别门限,并对样本量、判别精度及方法的合理性等相关问题进行了讨论。实例仿真取得了较好的判别结果,结果表明该方法对数据没有分布要求,简单可行。

Aiming at the distinguishing problem of the small samples whose distribution is unknown, a new extrapolated distinguishing model for gross error based on GM （1,1） model is proposed. The distinguishing principle and its approach are introduced. The distinguishing limits is ascertained based on the precision of GM （1,1） model and prediconted value. And the correlative problems, including the quantity of samples, the distinguishing precision, and the rationality of the method are discussed also. The simulation example results show that the method has no demand for the distribution of data and is feasible and effective.