Neural network system for the analysis of transient phenomena onboard the DEMETER micro-satellite
Abstract
In 2001, the DEMETER micro-satellite will belaunched to perform Detection of Electro-Magnetic EmissionsTransmitted from Earthquake Regions. Its main scientific ob-jective is related to the investigation of the ionospheric perturba-tions due to the seismic and volcanic activity. A system allowingan onboard identification and characterization of spatially andtemporally coherent structures associated with the measurementof one or several electromagneticwavefield components is used.It is based on neural networks. The choice and trainin gof theneural network are done on the ground from availablewaveforms.The parameters of the neural network system are then transmit-ted to the satellite. This reconfiguration process can be repeatedwhenever necessary durin gthe space mission. Details about thefunctionin gand codin goptimization for DSP implementation ispresented. The first function of this system which will be per-formed on the satellite DEMETER is the real-time identifica-tion and characterization of whistler phenomena. An applicationto the analysis of such phenomena observed in data from theAUREOL-3 satellite is exposed.