A novel flight maneuver recognition method based on distributed extended belief rule-based system

The accuracy of flight maneuver recognition will significantly affects the trajectory prediction, threat assessment and intent recognition in air combat situation analysis. However, existing methods face great challenges, such as high model complexity, weak capability in handling uncertainty and poor interpretability of result. To this end, a novel flight maneuver recognition method based on distributed extended belief rule-based (EBRB) system is proposed.First, the flight maneuver is considered as a combination of horizontal and vertical maneuver based on the concept of motion decomposition. On this basis, the maneuver characteristics for each maneuver plane are determined. Next, a distributed EBRB generation method based on Relief-F and IGR is developed to implement key characteristic selection and distributed EBRB generation. Finally, a maneuver rule search and activation method integrating the importance knowledge of maneuver characteristics and the ER algorithm is designed to achieve the inference of maneuver rules and obtain the maneuver recognition results. Experiments were carried out to validate the effectiveness and performance of the proposed method using the flight dataset from aircombat simulation.