Abstract: The multi-maneuvering target tracking(MMTT) in dense multi-return environments is studied.The properties of the joint probabilistic data association(JPDA)are analyzed,and data association is reduced to a sort of constraint combinatorial optimization problem. By using the combinatorial optimization strategy of the stochastic neural networks,a new fast adaptive neural network method for MMTT(FANJPDAF) is presented and its computation burden is re-duced drastically Simulation results show that the FANJPDAF is of high accuracy,good asso-ciation performance and tracking effectiveness.

Key words: multiple target tracking, data correlation, neural nets