基于无迹变换的协方差交集算法研究

Research of Covariance Intersection Algorithm Based on Unscented Transformation

针对分布式数据观测系统中数据融合存在数据线性化过程误差较大、滤波过程中的错误无法修正的问题,提出了一种基于无迹变换的协方差交集算法。该算法首先对系统数据进行无迹变换,然后对数据采用协方差交集算法滤波,可得到较好的滤波性能。实验结果表明,该算法弥补了协方差交集算法的不足,能够修正数据线性化误差及滤波产生的错误,是一种有效的数据融合算法。

In view of problems existed in data fusion of distributed data observation system, namely bigger error in data linearization process and uncorrecting error in filtering process, the paper proposed a eovariance intersection algorithm based on unscented transformation. The algorithm makes unscented transformation for system＇s data at first, then uses covariance intersection algorithm to filter data, which can gain better filtering performance. The experiment result showed that the algorithm makes up deficiencies of covariance intersection algorithm and can correct errors produced in processes of data linearization and filtering, which is an effective data fusion algorithm.