基于模糊推理的改进的交互式多模型算法

摘要

To improve the tracking precision of interacting multiple model algorithm,an optimization method for the interacting multiple model algorithm based on fuzzy inference was proposed.Model probabilities were used as input variables,and a coefficient of system noise covariance was used as the output variable for adjusting the system noise covariance and reducing the error.Taking the application of interacting multiple model filters based on Kalman filters as an illustration,the influence of system noise covariance on the tracking result was analyzed under the model matching and mismatching situations,and the interacting multiple model algorithms before and after improvement were compared.The results show that the interacting multiple model algorithm improved by fuzzy inference has better tracking performance.%为了提高交互式多模型算法的跟踪精度,提出一种利用模糊推理对交互式多模型算法进行优化的方法,以模型概率为输入,输出值为模型系统噪声协方差的系数,通过该系数来调节模型系统噪声协方差以减少误差;以基于卡尔曼滤波器的交互式多模型滤波器的应用为例,分别对交互式多模型算法在模型匹配和不匹配的情况进行实验分析,并对改进前、后的交互式多模型算法进行比较.结果表明,基于模糊推理的改进的交互式多模型算法能取得更好的跟踪效果.