A neural network model of the relativistic electron flux at geosynchronous orbit
Abstract
A neural network has been developed to model the temporal variations of relativistic (greater than 3 MeV) electrons at geosynchronous orbit based on model inputs consisting of ten consecutive days of the daily sum of the planetary magnetic index Sigma Kp. The neural network (in essence, a nonlinear prediction filter) consists of three layers of neurons, containing 10 neurons in the input layer, 6 neurons in a hidden layer, and 1 output neuron. The output is a prediction of the dailyaveraged electron flux for the tenth day. The neural network was trained using 62 days of data from 1 July 1984 through 31 August 1984 from the SEE spectrometer on the geosynchronous spacecraft 1982019. The performance of the model was measured by comparing model outputs with measured fluxes over a 6year period from 19 April 1982 to 4 June 1988. For the entire data set the RMS logarithmic error of the neural network is 0.76 and the average logarithmic error is 0.58. The neural network is essentially zerobiased, and for accumulation intervals of three days or longer the average logarithmic errors is less than 0.1. The neural network provides results that are significantly more accurate than those from linear prediction filters. The model has been used to simulate conditions that are rarely observed in nature, such as long periods of quiet (Sigma Kp = 0) and ideal impulses. It has also been used to make reasonably accurate dayahead forecasts of the relativistic electron flux at geosynchronous orbit.
 Publication:

NASA STI/Recon Technical Report N
 Pub Date:
 April 1991
 Bibcode:
 1991STIN...9210030K
 Keywords:

 Earth Orbital Environments;
 Electrons;
 Flux (Rate);
 Geosynchronous Orbits;
 Mathematical Models;
 Neural Nets;
 Relativistic Particles;
 Daytime;
 Neurons;
 Nonlinear Filters;
 RootMeanSquare Errors;
 Temporal Distribution;
 Astrodynamics