SAS glimmix cpoy算法拟合多水平log-binomial模型报错

利用copy算法后模型报错，Obtaining minimum variance quadratic unbiased estimates as starting values for the covariance parameters failed。无法获得协方差参数起始值的最小方差二次无偏估计

使用SAS glimmix拟合二水平的log-binomial模型，
自变量全部为分类变量，二水平设置为研究对象的生活地区（华北、东北、华南等5个地区），拟合log-binomal模型时显示不收敛。使用COPY算法，原始数据集weight=0.999，新数据集weight=0.001。重新拟合Log-binomial，显示：Obtaining minimum variance quadratic unbiased estimates as starting values for the covariance parameters failed

代码如下：
DATA ao ;
SET a ;
W=0.999;
RUN;
DATA ac ;
SET a ;
y= 1-y;
W=0.001;
RUN;
DATA aoc ;
SET ao ac ;
RUN;
proc glimmix data=aoc;
class motherage(ref=first) fatherage(ref=first) bmi(ref=first) hisabortion(ref=first) address;
weight w;
model livebirth(event="1") = motherage fatherage bmi hisabortion /solution oddsratio cl dist=bin link=log;
random intercept/sub=address;
run;
报错：Obtaining minimum variance quadratic unbiased estimates as starting values for the covariance parameters failed