基于DS算法的雷达目标识别方法研究

主要目的是为解决干扰存在下不同类型传感器、不同格式信息之间的融合问题,设计了一种较为有效的融合算法,来对敌方的危险目标进行识别;主要方法是把神经网络改进的BP算法与Dempster-Shafer(D-S)证据理论相结合,将来自于各种传感器探测设备多次观察所得到的数据,经过神经网络后,得到基本概率附值,然后利用DS证据理论进行实时的时域和空域融合,从而达到准确的目标识别;仿真结果表明该算法在有效提高识别概率的基础上,大大提高学习速度,结果可行.

Algorithm of Recognition for Radar Target Based on D_S Evidence Theory

For solving the problem of information fusion between different type of sensors in interference environment,a kind of more effective fusion algorithm is designed to make a more accurate identification of the dangerous target of the enemy.And the main method is using improved BP theory,Dempster-Shafer(D_S) evidence theory to fusion the data which come from multisensor via neural network to receive Basic Probability Assignment Function(BPAF),then using D_S evidence theory to fusion the data,thus achieves the accurate target identification.The simulation result showed: it does not only greatly improve the target recognition accuracy,but also improve recognition speed.The result was practicable.