Underwater target maneuver detection algorithm based on width learning

For underwater target tracking in the process of data sample size less problems, combining with the characteristics of the sample quantity required width of lower learning network this paper proposes a new mobile detection algorithm, existing maneuver detector algorithm in the test statistic to construct the network characteristic vector, training, through the network to network of incremental learning method reduces the detection delay. Simulation results show that the algorithm can calculate more accurate network weight coefficients after training with a given simulated track, and can give the possibility of the target in maneuver mode and non-maneuver mode in real time during tracking, which greatly improves the real-time performance of maneuver detectionn high-velocity projectile penetration into underwater targets.