A building deformation prediction method based on a non-parametric regression model

Abstract

The application discloses a building deformation prediction method based on a nonparametric regression model, and comprises the following steps: 1, obtaining the measured time series data of a building monitoring point, and performing noise reduction pretreatment on the data by using a sliding average method; 2, establishing a nonparametric regression model for the pretreated data, and performing autocorrelation test on irregular terms in the nonparametric regression model; 3, performing parameter optimization on SVR by using a fish-eagle optimization algorithm, and then establishing an SVR irregular term regression prediction model by using optimal parameters to fit irregular terms; 4, obtaining a final nonparametric regression model; and 5, performing deformation prediction on the building. The fish-eagle optimization algorithm can obtain a global optimal solution of an optimization problem, the optimization result is independent of initial conditions and the algorithm is independent of a solution domain, has strong robustness, can be well used for solving model parameters, and the method can realize accurate deformation prediction.

Images