Title :
Obtaining the real height profile of ionospheric electron density with BP network
Author :
Qinglin, Wu ; Xianwen, Li ; Yan, Wu ; Chong, Wu
Author_Institution :
Coll. of Electron. Inf., Wuhan Univ., China
fDate :
6/21/1905 12:00:00 AM
Abstract :
Multi-layer feed forward neural networks can approximate any nonlinear mapping. Because of this nature it is used to solve inverse problems. When we apply BP network to inverse the real height profile of ionospheric electron concentration from the ionogram obtained by vertical sounding, the maximum relative error is less than 1%.
Keywords :
backpropagation; feedforward neural nets; geophysics computing; inverse problems; ionospheric techniques; plasma density; BP network; inverse problems; ionospheric electron concentration; ionospheric electron density; maximum relative error; multi-layer feed forward neural networks; nonlinear mapping; real height profile; vertical sounding; Educational institutions; Electrons; Feedforward neural networks; Feeds; Frequency; Integral equations; Ionosphere; Linear approximation; Multi-layer neural network; Neural networks;
Conference_Titel :
Computational Electromagnetics and Its Applications, 1999. Proceedings. (ICCEA '99) 1999 International Conference on
Print_ISBN :
0-7803-5802-3
DOI :
10.1109/ICCEA.1999.825138
