论文部分内容阅读
本文提出了一种自适应进行模型修正的变结构交互式多模型(DMM VS-IMM)算法。在该算法中,道路信息的利用体现在两方面:一是利用道路对目标运动的约束修正滤波使用的运动模型,包括对状态转移矩阵的实时修正和对过程噪声的修正。二是利用道路信息实现变结构交互式多模型滤波,即利用预测的状态估计值和道路信息判断目标当前时刻在道路网中所处的位置,从而相应地调整模型集中模型的数量和种类,然后完成多模型滤波。该算法将模型修正和变结构交互式多模型滤波有效地结合,使目标在单条道路上和道路交叉点处的状态估计精度都得到改善。蒙特卡洛仿真结果验证了该方法的有效性及鲁棒性,且其计算量适中。
This paper presents a variable structure interactive multi-model (DMM VS-IMM) algorithm that adapts to model modification. In this algorithm, the utilization of road information is reflected in two aspects. One is to use the constraint model of road motion correction to the target motion, including the real-time correction of state transition matrix and the correction of process noise. The second is to make use of the road information to realize the variable structure interactive multi-model filtering. That is to say, using the predicted state estimate and road information to determine the current position of the target in the road network, the number and types of models are adjusted accordingly Complete multi-model filtering. The algorithm effectively combines model correction and variable structure interactive multi-model filtering to improve the state estimation accuracy of the target on a single road and at a road intersection. Monte Carlo simulation results verify the effectiveness and robustness of the proposed method and its computational complexity.