病态加权总体最小二乘靶向病灶的正则化方法

病态EIV模型的病灶源于设计矩阵的部分数据列之间存在复共线性关系。针对病灶特点制定正则化策略，在克服病态性的同时尽量减小正则化过程所引起的副作用，提出靶向病灶的正则化方法。通过数值试验，与总体最小二乘方法、病态总体正则化方法等进行比较，结果表明靶向病灶的正则化方法最优。

Target Focus Regularization as Compared with Ill-Posed Weighted Total Least Square

Usually in ill-posed error-in-variable models, only part of columns in the design matrix possess multicollinearity. This is where the model focuses. According to the focus characteristics, the corresponding regularization strategy is made, which can overcome the ill-posed effects and minimize side effects caused by the process of regularization at the same time. We propose target focus regularization as opposed to the ill-posed error-in-variable model. Numerical experiments show the proposed regularized total least-squares solution method performs better than others.