一种新的航迹与量测互联技术

为了同时解决存在于航迹与量测互联问题中的两种不确定性：目标运动和量测源的不确定性，本文提出一种基于BP的变结构交互式多模型2-D互联技术。首先，采用全局2-D分配算法确定量测与目标的分配问题，确定量测序列；其次，提出一种元模式的概念，采用离散小波变换，将目标量测序列映射到目标模式序列，从而由BP神经网络对目标运动模型对应的概率进行估计。再利用变结构的交互式多模型进行目标滤波估计。初步实验结果表明：本文提出的技术能较好地解决密集目标、特别是机动目标的互联问题。

A novel track-to-measurement association technique

2-D track-to-measurement association approach, a BP based Variable Structrue Interaction Multiple Model (IMM), is proposed to simultaneously resolve the two most important uncertainties existed in the data association problem, i.e. measurement origin uncertainty and target motion uncertainty. After deciding the measurement-to-track correlation via global 2-D assignment, a discrete wavelet transform is used to map the measurement sequences to meta-mode sequences, then the probability of the target motion model is estimated through BP neural network, and target filtering and estimatation are completed via a variable structure IMM estimator. Simulation results show that the proposed technique can solve track-to-measurement association, especially maneuvering target association problem properly.