论文部分内容阅读
UKF算法是广泛应用的非线性滤波方法之一,在加性噪声条件下,根据是否状态扩展和是否重采样有四种实现方式.从算法精度、适应性和计算效率等方面进行了理论分析和仿真计算,证明适当选择滤波器参数,常用采样策略下,状态扩展与非扩展的UT变换结果相同,但后者的计算效率较高;加性测量噪声条件下,扩展与非扩展UKF可获得相同的滤波结果;加性过程噪声条件下,扩展与非扩展UKF仅能获得相同的状态预测结果;重采样不总是构建滤波器的必要环节,但理论分析和仿真计算发现了重采样对滤波器增益的自适应调节能力,指出其在状态偏差或未知机动模式较大时对改善滤波器收敛性和精度有重要贡献.在此基础上,给出了实际应用中的滤波器设计准则:对于加性测量噪声宜采用非扩展方式;对于加性过程噪声,状态偏差或机动较小时宜采用扩展或非扩展的非重采样方式,系统状态偏差或机动较大时宜采用非扩展的重采样方式.性能分析方法和系统设计准则对研究其它的滤波器同样有参考价值.
The UKF algorithm is one of the widely used nonlinear filtering methods. Under additive noise conditions, there are four ways to implement the UKF algorithm based on whether the state is extended or whether it is resampled. The theoretical analysis and calculation of the algorithm are carried out in terms of accuracy, adaptability and computational efficiency Simulation results show that the parameters of the filter are properly selected. Under the common sampling strategy, the results of UT transform with state extension and non-extension are the same, but the calculation efficiency of the latter is high. Under additive measurement noise, the same can be obtained for extended and non-extended UKF The results of the filtering process are the same with those of the UKF in the case of additive process noise. The resampling is not always necessary to construct the filter. However, the theoretical analysis and simulation results show that the resampling filter Gain adaptive adjustment capability, and pointed out that it has an important contribution to improving the convergence and accuracy of the filter when the state deviation or the unknown maneuvering mode is large.On the basis of this, the design criterion of the filter in practice is given: Sex measurement noise should be non-expansion mode; for additive process noise, the state deviation or maneuver should be less expansion or non-expansion should be non-resampling mode, Non-extended resampling should be used when the system state deviation or maneuvering is large.Performance analysis methods and system design guidelines are equally valuable to the study of other filters.