Research On Classification Algorithm Of Multivariate Time Series Data Based On TLE Dataset

Spacecraft technology is one of the most influential science and technology in the modern world.However,the space debris generated by human space activities has caused great security threats to the spacecraft being used,and the the spacecraft of non-cooperative forces also pose a threat to their own spacecraft.Therefore,it is more and more important to identify the spatial object.The usual method is to classify and identify the spatial object.The current classification and recognition methods of space target are mainly based on static characteristics,while in this paper,we propose a method applying multivariate time series data mining to study the historical data of spatial object movement,which can be used to classify the spatial objects.Based on the concept of velocity and acceleration in physics,we introduced it into multivariate time series data problems.In this paper,we propose a method(Velocity-Acceleration feature Extending,VAE)of velocity-acceleration feature expansion for extending the variation features of the original time series.The multivariate time series classification algorithm based on VAE method(Velocity-Acceleration feature Extending based Time Series Classification algorithm,VAE-TSC)is proposed,and the effectiveness of VAE-TSC algorithm is verified by experiments.In view of the shortcomings of the current methods of extracting local sequence features,this paper puts forward a new segmentation method based on maximum and minimum(Maximum-Minimum segmenting MM).This method is not only a universal method,and the local temporal feature extracted from the obtained subsequence also has a clear meaning.Combining the MM method and the VAE method,the VAEMM method is formed.And the method of multivariate time series classification algorithm based on VAEMM(Velocity-Acceleration feature Extending and Maximum-Minimum segmenting based Time Series Classification algorithm,VAEMM-TSC)is put forward,and the effectiveness of VAEMM-TSC algorithm is verified by experiments.Based on the aforementioned research in this paper,the classification algorithm based on the combination of dynamic features and static features(Static-Dynamic-Combined classification algorithm,SDC)is proposed for the classification of space targets.By contrast experiments,the validity of SDC algorithm is proved.