AI-Driven Space Debris Detection and Trajectory Prediction for Enhanced Safety at Bhartiya Antriksh Station

Space debris is a critical threat to operational assets, including the Bhartiya Antriksh Station (BAS). This paper presents a real-time system for debris detection and trajectory prediction using Two-Line Element (TLE) data. The methodology consists of two main phases: detection and prediction. In the detection phase, TLE data is preprocessed to extract key orbital parameters such as inclination, eccentricity, and mean motion. A Support Vector Machine (SVM) classifier is employed to differentiate between operational satellites and debris. Additionally, a neural network technique is implemented to capture nonlinear patterns, thereby enhancing detection accuracy. In the prediction phase, historical TLE data is analysed to forecast future debris trajectories. An Autoregressive Integrated Moving Average (ARIMA) model is fine-tuned using grid search, information criteria, and residual diagnostics to serve as a baseline for time-series prediction. Complementing this, a Long Short-Term Memory (LSTM) network is utilized to capture complex temporal dependencies and improve forecast precision. Experimental evaluations indicate that the integrated system reliably detects debris and accurately predicts its motion, providing timely alerts for collision avoidance. The proposed method establishes a robust foundation for advanced, real-time debris management and contributes to the operational safety and sustainability of BAS.