基于修正卡尔曼滤波的目标跟踪

分析了卡尔曼滤波算法在目标状态发生突变和运动模型建立不准确时估计精度降低，甚至发散的原因，对比自适应渐消卡尔曼滤波算法，提出了一种通过直接修正预测值来提高卡尔曼滤波算法精度、改善算法性能的修正算法。修正的算法通过设置判定准则和修正准则，实时修正预测值，在滤波初始阶段可迅速降低估计误差、提高稳态时的滤波精度、缩短收敛时间；当目标发生状态突变时，可消除或降低由于目标状态突变造成的滤波跟踪精度下降、滤波发散的问题；当目标运动建模不准确时，可消除或降低由于建模不准确带来的模型误差。仿真实例说明了算法的有效性和较强的实际应用指导意义。

Target tracking based on amendatory Kalman filter

The reasons why Kalman filter’s estimation precision is low and the filter is divergent when the target moving state changes or the moving model is not accurately established are analyzed. In contrast with the adaptive fading Kalman filter, a new amending algorithm is put forward which improves Kalman filter estimation accuracy and performance by adjusting the predicted value directly. The amending algorithm adjusts the predicted value in time by setting judgment and amendment rules, which can quickly reduce estimation error in the initiating filter stage, increase the precision in the static filter stage and shorten the convergence time. The new algorithm is able to reduce or eliminate the situation where the tracking precision declines and the filter is divergent when the moving state changes. Also, it can reduce or eliminate the model error when the moving model is not accurately established. The results of simulation indicate the effectiveness of the new algorithm and its strong practical guiding significance.