Orbital Maneuver Fast Detection Of GEO Satellite Based On Artificial Intelligence

Geosynchronous orbit(GEO)satellite has unique stationary characteristics,so it plays a vital role in the space situational awareness syste.The orbit determination and maneuver detection of GEO satellite is the basis for real-time monitoring of the operation status of GEO satellite and ensuring the safe and stable operation of the satellite.In recent years,many countries have been deepening the exploration of space,resulting to space goals’ increasing and frequent space activities,which have put forward higher requirements for the space situational awareness system.GEO satellite has high orbit and frequent mobility,which brings difficulties to the satellite orbit determination and maneuver detection.At the same time,artificial intelligence technology has developed rapidly in recent years,and is widely used in various fields such as industry,finance,medical care and other fields.Artificial intelligence technology can accurately extract features and accelerate the speed of information processing,which can solve the current problems of large error and low recognition rate in satellite orbit determination and maneuver recognition.Therefore,this paper takes GEO satellite as the research goal,and applies the artificial intelligence technology to the field of space situation awareness.According to the characteristics of GEO satellite orbit data and the advantages of artificial intelligence technology,it proposes the orbit determination and maneuver detection method of GEO satellite based on artificial intelligence.The deep neural network,extreme gradient boosting(XGBoost)algorithm and long short term memory(LSTM)network are used to realize the fast orbit determination and maneuver detection of GEO satellite.The main research contents are as follows:(1)A GEO satellite orbit determination method based on deep neural network is proposed.The deep neural network can extract the relationship between the GEO satellite position sequence measured by the ground base station and the corresponding orbit parameters,and predicts the satellite orbit parameters according to the position sequence of the satellite operation,and the final predicted value is basically consistent with the true value.The average error of eccentricity is 0.026,the average error of orbital inclination is 1.633°,and the average error of longitude of ascending node is 1.794°.(2)A GEO satellite trajectory prediction and maneuver detection method is proposed,which used XGBoost algorithm to predict the GEO satellite distance difference sequence measured by ground base station.The average error of the trajectory prediction is less than 0.11 m under different lengths of baseline.According to the prediction results from baselines in two directions,the detection of satellite maneuver is realized and validated on GEO satellite.The overall identification accuracy of satellite maneuver is 77.78%.(3)A method of GEO satellite trajectory prediction and maneuver detection based on LSTM is proposed,establishes a suitable LSTM network model,and predicts the distance difference of GEO satellite measured by the ground base station under different lengths of baseline.the average error is less than 0.14 m.Based on the prediction results from baselines in two directions,the detection of GEO satellite maneuver is realized and verified on the GEO satellite data,and this method can achieve 92.22%accuracy when the baseline reaches to 2 km.Compared with the XGBoost model,the trajectory prediction error of this method is slightly higher,but the maneuver detection accuracy is higher.