采用方差模型概率的机动目标跟踪算法

摘要

针对交互式多模型（ IMM ）算法计算量大、模型切换时性能不佳的特点，提出了一种新的机动目标跟踪算法---方差模型概率（ Variance Model Probability ，VMP）算法。该算法结合多模型思想，利用当前量测残差在线推导模型方差，自适应调整模型概率。模型概率大小与方差成反比，滤波输出为各模型加权和。为减小量测噪声引起的误差影响，在设定的时间窗内求方差平均值。仿真结果表明，VMP算法不仅性能优于交互式多模型算法，同时也减少了计算量，提高了费效比。%The interacting multiple model ( IMM ) algorithm has large amount of calculation and limited performance during model switching .To solve the problem,a new adaptive filtering algorithm,Variance Model Probability ( VMP) algorithm,was introduced .The algorithm uses the multiple models and deduces model variance by using measurement residuals .The model probability has inverse relationship with variance,and the filtering result is the weighted sum .In order to reduce measurement error caused by noise,it needs to calculate average variance in set time window .Simulation indicated that VMP algorithm is better than IMM in performance and needs less time in calculation .It can improve the cost-effectiveness effectively .