Abstract:The standardized residuals of polluted semi-parametric models have been influenced by model system structures when making use of the method of quasi-accurate detection of gross errors, so it is difficult to choose the quasi-observations in the primary selection in this way. In this paper, the observations are selected with the modulus maxima under the wavelet transform filter analysis as primary quasi-observations. Then the location and value of gross errors are realized on condition of the minimum of the real errors related to quasi-accurate observations. At last, penalized least square estimation is used to separate the systematic errors to obtain parameter estimation accurately.
DENG Xiaodong,ZHANG Xianyun,ZHANG Jun et al. Resolution of Polluted Semi-Parameter Model with Wavelet Assisted Quasi-Accurate Detection and Penalized Least Square Estimation[J]. jgg, 2016, 36(5): 442-.