Correlation and Forecasting Analysis on Lost Satellites.

摘要

The relationship between the size of the United States Air Force Space Command's lost satellite lists and thirteen other variables is investigated. The thirteen variables contain the Space Surveillance Center crew's effort to work the lost list, solar activity, geomagnetic field strength, and the number of observations received each day from seven space track sensors. To identify the relationship of cause and effect between the lost lists and the thirteen variables, simple and multiple linear regressions are used on a data set that begins on 1 January 1988 and ends on 31 December 1988. Multiple linear regression and basic time-series smoothing techniques are used to forecast the number of lost satellites. The results of this study show that solar activity causes near earth satellites to 'go lost'. This study also shows that not all sensors respond to increases in the size of lost lists in the same manner. Finally, the best forecast for the size of the lost satellite lists is provided by a multiple linear regression. (jhd)