A Missing Values Imputation Method for Radiation Data from Beidou MEO Satellite Sensors

Radiation dose refers to the sum of ionization energy deposited by various incident high-energy particles in unit mass material. The radiation dose data detected by Beidou MEO satellites suffers from the problem of missing values that tend to occur continuously and there is little difference in the radiation dose detection values in the three directions in the satellite cabin. Firstly, the detector 3 data with strong regularity is selected as the research object. In order to fill the missing values, a sine-adjusted linear spline regression method is proposed, in which each continuous data-missing time period is individually processed in a spline and the missing values are filled with the sine-adjusted linear equation. The experiment shows that with our approach the filling curve is highly consistent with the true curve in terms of growth trend and periodic fluctuation. And our method performs significantly better with respect to errors and correlation coefficient between true and filled values, than alternative methods such as forward interpolation and linear interpolation. This paper selects the data of different time periods to analyze the filling effect, and draws the above conclusions. The proposed method properly solves the problem of continuous data-missing, leading to a data set with good completeness and high accuracy, which lays a foundation for the subsequent tasks like data release, modeling and visualization.