列车组合定位的改进UPF算法探讨

Study on Improved UPF Algorithm to Train Integrated Positioning

针对传统粒子滤波缺乏当前量测信息、容易出现粒子退化现象、滤波精度不高,难以应用于GNSS/INS列车组合导航的问题,提出了一种改进粒子滤波算法。通过将无迹卡尔曼滤波框架应用到标准粒子滤波中,产生粒子的重要性函数,考虑了当前量测对状态估计的影响,改善了滤波效果。在重采样环节又融入了马尔科夫链蒙特卡洛方法,增加了采样粒子的多样性,提高了滤波的精度。结合采集某列控系统的样本数据进行仿真,结果表明:改进的UPF与传统的UPF相比,滤波效果更好,定位精度更高,在GNSS/INS列车组合定位中有更好的工程使用价值。

In the traditional particle filter,the lack of measurement information is easy to appear the phenomenon of particle degradation,the filtering accuracy is not high,which is difficult to be applied to the GNSS/INS train navigation,thus an improved particle filter algorithm is put forward.By applying unscented Kalman filter framework to standard particle filtering,the importance function of particle is produced.Considering the current measurement of state estimation,the effect of filtering is improved.By integrating the resampling method into the way of Markov Chain Monte Carlo,the diversity of the particle is increased,and the filter precision is improved.The sample data of a certain train control system is simulated.The results show that compared with the traditional UPF,improved UPF has better filtering effect and higher positioning accuracy,the value of the project is better in the combination of GNSS/INS train.