Spatial Non-Cooperative target behavior intent recognition based on data generation and deep neural networks_dataSet

Under the condition of informatization, the space environment has become increasingly complex, and the number of Non-Cooperative targets in space is increasing, and it is difficult for ground operators to quickly and accurately identify their intentions according to the movement laws of Non-Cooperative targets. Therefore, this paper proposes a spatial Non-Cooperative target behavior intent recognition model based on Stacked Autoencoder (SAE) and Gated Recurrent Network (GRU) to assist ground operators in identifying the intention of Non-Cooperative targets. Firstly, the autoencoder is used to compress the time series data and extract the key features. Subsequently, the GRU network was used to classify the trajectories. At present, there is no public orbital data of Non-Cooperative target behavior available, and it is difficult to fully train the model by relying only on a small amount of known data. In order to solve the problem of poor recognition effect caused by insufficient samples, this paper proposes a simulation sample generation method, through which a large number of orbital data of target behavior are obtained through simulation for the identification of the behavior intention of space Non-Cooperative targets. Firstly, by using the J2 invariant condition, through the analysis of approach intention, detection and orbit keeping intention, the relative position between the Non-Cooperative target and the target satellite and the relationship between the orbit root number are obtained, and then the orbit root number of the Non-Cooperative target and the target satellite is obtained. Considering the orbit prediction model of J2 perturbation, the orbit data of various intentions are derived. After obtaining the simulation data, the experimental results show that compared with the single models such as Long Short-Term Memory Network (LSTM), GRU-FCN, SAE and Backpropagation (BP), the proposed method has a significant improvement in accuracy and loss value performance indexes, reaching an accuracy of 97.8%.