Adaptive nonlinear GM-PHD filter and its applications in passive tracking

Firstly, to solve the nonlinear problem in the field of passive tracking, Gauss-Hermite quadrature is used to Gaussian mixture probability hypothesis density filter, and the quadrature Kalman probability hypothesis density filter was proposed. Then under the condition of unknown and time-varying process noise statistic, a noise statistic estimator based on maximum a posterior estimation was used in probability hypothesis density filter. According to the residual between predicted state and estimated state, an algorithm to judge and restrain filter divergence was proposed. Finally, simulations under the condition that two passive sensors tracking multiple targets show that:the proposed algorithm has better accuracy than existing algorithms, and achieve good effect when process noise statistic is unknown and time-varying.