N步返回可能模型集算法比较研究

A Comparative Study on N-step Back Likely-model Set Algorithm

可能模型集算法由于不能保证模型集转换的正确和及时,给目标的状态估计带来较大的误差。将可能模型集算法返回n步,以保证及时初始化新激活的模型,能在一定程度上降低状态估计的误差,这同时也导致计算量的增加。文中针对多种不同返回步数的可能模型算法进行了比较分析,仿真结果表明,算法返回一步或两步能够大大提高目标状态估计的精度,尤其是当使用的模型集规模较小时。

Since likely model set（LMS） algorithm can＇t guarantee that model set adaptation is done correctly and timely,errors may be introduced in the target state estimate.If LMS go back n-step in time to initialize the newly activated models,the state esti-mate error will be reduced in some degree.But this method also increases the computation complexity.Some different N-step back LMS algorithms are compared.Simulation results indicate that LMS go back one or two steps can improve the precision of the state estimate,especially when the model set used is small.