单输入单输出回归模型的异方差性检验

在研究木材材质特性时,通常将木材本身看作一个系统,输入不同的影响因素,如含水率、密度、节子、腐蚀等会得到不同的机械力学特性,而相关的回归分析一直是在对上述各测试观测值的方差齐性默认的情况下进行的,并未对其方差异性进行过深入研究．本文研究基本系统的单输入单输出回归模型的方差齐性检验问题．通过推导得到似然比检验统计量与Score检验统计量,并且基于参数的正交化得到修正的似然比检验统计量与修正的Score检验统计量,并利用Monte-Carlo方法模拟了几种统计量的功效,结果显示,这几种检验统计量都有很好的检验效果．

The test of homogeneity of variance of single-input-single-output control model

In the study of the wood properties,wood is usually treated as an input-output system,for example,when different influence factors such as moisture content,density,knots,corrosion level are input,different mechanical properties could be got.However,the regression analysis is developed under the default homogeneity variance of the tested value,and there is no intensive study of the homogeneity variance.The paper focuses on the test of homogeneity of variance of single-input-single-output(SISO) control model.The likelihood ratio test statistic and the Score test statistic were derived,and then revised by parameter orthogonalization.The efficiency of the test statistics was stimulated by Monte-Carlo method.The results show that they all have good effect.