D-S证据理论在目标识别中的应用

根据地面目标运动引起的地震动信号的特征信息，应用多传感器信息融合的方法将目标正确分类。首先根据地震动信号在频域和时频域的多种特征，应用BP神经网络模式识别法，将地面车辆目标分为轮式车、轻型履带式车和重型履带式车。设计了一种以神经网络正确识别率作为基本概率赋值的方法，并应用D－S证据理论进行识别信息融合，训练样本和识别样本分别取自外场实验所获得真实有效的数据，通过对识别信息融合，以较高的可信度得到与识别样本相一致的识别结果，这表明所设计的获取基本概率赋值的方法及信息融合算法是有效的，该方法可以推广应用于其他多传感器或多信息源的探测识别系统中。

THE APPLICATION OF D-S PROOF THEORY IN TARGET IDENTIFICATION

It is discussed in this paper that the targets moving on the ground are identified according to the seismic signal by the way of multi sensor information fusion. Based on the features of seismic signal in frequency domain and time frequency domain, the BP neural network is used to classify the target into wheeled vehicle, light tracked vehicle and heavy tracked vehicle. A method is proposed which regards the identification rate of the BP neural network as the basic probability value. The identification rates are fused and the result is obtained with the high reliability and which is in accordance with the original sample. It is shown that the method proposed in this paper is valuable, and it can be broadly used in the other analogous system.