Discussion On Some Problems Of Semi-parametric Model

The semiparametric model adds the nonparametric component to the observation equation.On the one hand,the mathematical model is closer to the real situation,and the other is the numerical error and the error of the accidental error.In this paper,we discuss several problems of the semiparametric model,discuss the combination of several traditional models and semiparametric models,and analyze the difference between the least squares criterion and the least squares criterion,and discuss several data processing methods and semiparametric And the validity of the parameters obtained by the combination of the traditional model and the semiparametric model is obviously improved by using the example.The main research contents are as follows:1.The semiparametric model and the compensated least squares estimation method are introduced,and the selection of normal matrix and smoothing factor are summarized.The stability of the traditional parameter model and the semi-parametric model for the system error is compared and analyzed.Aiming at the problem that the traditional smoothing factor selection method is complex and computationally large,a new method for selecting the smoothing factor-improved efficiency method is proposed.Smoothing factor selection method.2.The gray prediction modeling method based on semi-parametric model is studied,and by improving the selection of the traditional normal matrix principle,The influence of different normal matrices on the parameter estimation and the precision variation are analyzed.Combined with the deformation monitoring data,the gray forecasting under the least squares criterion and the gray prediction precision under the least squares criterion are discussed.3.The semiparametric model and its ridge estimation are studied,and the solution of the model when the matrix is ill is discussed.The relationship between the ridge estimation of the semiparametric model and the least squares estimation of the pan compensation is studied.The method of selecting the parameters of the ridge is summarized and the influence of the semi-parametric model and the semi-parametric ridge estimation on the parameter estimation is analyzed by an example.4.Considering that the least squares estimation has no robustness,the compensated least squares estimator does not have robustness.The semiparametric model robust estimation is studied,and the selection method of the robust weighting factor is discussed.The accuracy of several model solutions and the influence on the accuracy are analyzed and compared by examples.