Application of SVM and PSO Arithmetic in Deep Space Exploration Data Analysis
A method of SVM optimized by using the PSO arithmetic is presented to solve nonlinear regression estimation problems in deep space exploration data analysis. This method is used to process the microwave brightness temperature (TB) data acquired by the CE-1 satellite. Firstly, the SVM regression model is established and some parameters of which are optimized by using the PSO arithmetic. Then, by training the TB data with the optimized SVM model, the relationship between the TB from four frequency channels and the lunar hour angle is established. Finally, the distribution maps of TB from four frequency channels on the entire lunar surface in certain short period are obtained. The error analysis indicates that the results of this paper can be used in the further study of lunar regolith depth. Furthermore, the abnormal data among the measured data can be found out and modified by using this method.
KeywordsSVM PSO TB data CE-1 Hour angle Data analysis
- 2.Du, S.-X., Wu, T.-J.: Support vector machines for regression. J. Syst. Simul. 15(11), 1580–1663 (2003)Google Scholar
- 3.Zheng, Y.C., Bian, W., Su, Y., et al.: Brightness temperature distribution of the moon: result from Chinese Chang’E-1 Lunar Orbiter. In: Goldschmidt Conference 2009, 21–26 June 2009 (2009)Google Scholar
- 7.Deng, N., Tian, Y.: New Method for Data Mining: Support Vector Machine, pp. 77–78. Science Press, Beijing (2006)Google Scholar
- 8.Xi, X., Wang, W., Gao, Y.: Fundamentals of Near-Earth Spacecraft Orbit, pp. 20–36. National Defense University Press, Changsha (2003)Google Scholar
- 9.Zhou, M.X., Zhou, J.J., Wang, F.: Analysis and simulation of microwave brightness temperature on lunar surface. In: 60th International Astronautical Congress (2009)Google Scholar