基于DS证据理论的多传感器信息融合算法

Multi-sensor information fusion algorithm based on DS evidence theory

多传感器卡尔曼滤波算法具有良好的信号跟踪及估值能力,但由于信号噪声的影响,不同传感器所提供的信号会产生一定幅度的偏差,不利于对真实信号的预测及估计。为了解决上述问题,提出了基于DS证据理论的多传感器量测融合的方法,利用证据理论处理传感器的量测信息以及滤波器的估计值,从而合理地计算出单传感器的权值,并对子传感器的量测值进行二次赋值。经过融合后的结果具有良好的滤波效果。将改进的量测融合方法应用到目标跟踪问题中,获得的目标跟踪精度提高了近一倍。通过仿真实验对比验证了新算法的可靠性及精确性,表明该方法具有一定的实用价值。

Multi-sensor Kalman filtering algorithm has good signal tracking and estimation ability.However,due to the influence of signal noise,the signals provided by different sensors would produce a certain level of deviations,and this will adversely influence the prediction and estimate to the true signals.In this paper,a method of multi-sensor measurement fusion was proposed based on the Dempster-Shafer evidence theory,in which the evidence theory was used to process the measurement information and the filter＇s estimate,thereby reasonably work out the weight of single sensor and make the second assignment to the sensor measurement values.The result after fusion has good filter effect.Then the improved measurement fusion method is applied to the target tracking problems,and the target tracking accuracy is almost doubled.The experiment results confirm the reliability and the accuracy of the algorithm,showing that it has certain practical value.