改进交互多模型的GNSS/航位推算列车组合定位算法

Improved interactive multi-model algorithm for GNSS/dead reckoning train integrated positioning

针对列车在实际运行环境下噪声先验信息未知或不准时导致滤波精度下降以及机动状态下模型建立与实际情况不匹配的问题,该文提出了一种基于H_(∞)滤波器改进的自适应交互多模型滤波(H_(∞)-AIMM)算法。该算法(1)在传统交互多模型(IMM)算法的基础上,采用误差压缩率之比来自适应调节模型转移概率;(2)将H_(∞)滤波器置于IMM算法前端得到噪声统计特性,进而对目标状态进行估计;(3)选取GNSS/航位推算列车组合定位系统作为实验背景,对H_(∞)-AIMM算法有效性进行了仿真验证。结果表明:所提算法相比H_(∞)-IMM算法和IMM算法定位精度分别提高了8.15%、19.65%,在列车发生机动前后的定位误差小,能有效提高目标定位精度,对于列车组合定位具有一定的工程应用价值。

Aiming at the problem that the noise prior information of the train is unknown or inaccurate in the actual operating environment,which leads to the decline of the filtering accuracy and the mismatch between the model establishment and the actual situation in the maneuvering state,this paper proposed an improved adaptive interactive multi-model filtering(H∞-AIMM)algorithm based on H∞filter.Firstly,based on the traditional interactive multi-model(IMM)algorithm,the algorithm adopted the ratio of error compression ratio to adaptively adjust the model transition probability.Secondly,the H∞filter was placed in the front end of the IMM algorithm to obtain the statistical characteristics of noise,and then the target state was analyzed.Finally,the GNSS/dead reckoning(DR)train integrated positioning system was selected as the experimental background,and the effectiveness of the H∞-AIMM algorithm was simulated and verified.The results showed that compared with the H∞-IMM algorithm and the IMM algorithm,the positioning accuracy of the proposed algorithm was improved by 8.15%and 19.65%,respectively,the positioning error before and after the train maneuvers was small,which could effectively improve the target positioning accuracy,and had certain engineering application value for train integrated positioning.