一种重尾量测噪声下的高超声速飞行器跟踪算法

A Hypersonic Vehicle Tracking Algorithm Based on Heavy Tailed Measurement Noise

为了解决在实际系统中因野值干扰带来的高超声速飞行器跟踪精度下降的问题,提出了一种交互多模型变分贝叶斯滤波算法(IMM-VB),该算法通过子模型权重与马尔可夫转移矩阵获取子模型的状态预测值。随后采用具有重尾特性的学生t分布取代高斯分布来描述量测模型,并利用VB算法实现子模型的量测协方差与状态的联合估计。最后在交互式多模型(IMM)框架下更新子模型权重与目标状态的融合输出。仿真结果表明,在野值观测条件下该算法比IMM算法具有更高的跟踪精度。

In order to solve the problem of degraded tracking accuracy of hypersonic vehicle caused by outliers disturbance in real systems, an Interactive Multi-Model Variational Bayesian （IMM-VB） filtering algorithm was proposed. The algorithm obtained the state prediction values of the sub-models through the sub-model weights and the Markov transition matrix. Then the student＇s t distribution with heavy tail characteristics was used to replace the Gaussian distribution to describe the measurement model and VB algorithm was used to realize the joint estimation of the measurement covariance and state of the sub-model. Finally, the fusion output of the sub-model weight and the target state were updated under the IMM framework. Simulation results show that the algorithm has higher tracking accuracy than IMM algorithm under outliers observation conditions.