Target trajectory prediction based on optimized neural network
Xiaoxiang Song, Yan Guo, Ning Li, Baoming Sun · 2017
In this paper, we propose a target trajectory prediction method based on optimized Neural Network. First, to make the trajectory prediction independent on the moving model of the target, we use Back Propagation (BP) Neural Network which has the complex nonlinear mapping ability and large-scale parallel distribution processing ability to analyze and predict target trajectory. Then, in view of the defects of Back Propagation Neural Network, such as slow convergence speed, easily to be trapped in local minimum and processing results will produce large error when the data has big fluctuation, an improved method is put forward. Simulation results show that the proposed model has obvious advantages in nonlinear fitting, convergence speed and prediction accuracy.