Lateral Maneuver Discrimination for Hypersonic Glide Vehicles: A Hybrid Approach Combining Model-Driven and Data-Driven Methods

Hypersonic glide vehicles (HGVs) are high-speed, highly maneuverable aerial platforms designed for near-space operations, posing significant challenges in detection, tracking, and trajectory prediction. This article introduces a hybrid approach that combines model-driven and data-driven methods to enhance lateral maneuver discrimination for HGV targets. The approach initially uses a model-driven state estimator to extract target states and lateral maneuver guidance parameters, i.e., the bank angle. Building upon this, a time-series-based neural network is designed for bank angle prediction and lateral maneuver discrimination. Specifically, we use an interactive multiple model (IMM) algorithm with varying process noise variances as a model-driven approach, based on the maneuver reentry vehicle (MaRV) model. In addition, we introduce a long short-time memory (LSTM) network with an attention mechanism as the data-driven method. Comparative analyses demonstrate the superior performance of the hybrid framework.