Neural network recognition of ionospheric anomaly events based on GNSS RO data

In order to develop the first generation satellite brain laboratory prototype, we have carried out a technology research in its intelligent cognition system. Specifically, based on GNSS (Global Navigation Satellite System) RO (radio occultation) data, the neural network cognition method of ionospheric anomaly events is established. For the three kinds of ionospheric anomaly events (sporadic E layer, ionospheric plasma bubble and space activity), by establishing input data samples and setting hidden layer, the recognition rates of neural network are 96.4
^%
, 98.3 % and 97.7%, respectively. This result proves the effectiveness of the neural network in ionospheric anomalies recognition. The next improvement aspects include increasing more kinds of physical quantities to the input data samples as well as adding the intensity cognition or classification module of ionospheric anomaly events.