基于多尺度Kalman滤波的多传感器数据融合

通过分析多尺度动态系统模型,提出了一种基于小波变换的Kalman多传感器数据融合算法。该算法结合了Kalman滤波的实时性、递归性和小波变换的多尺度特性,能对多传感器的观测数据有效地融合。算法首先将最细尺度上观测数据滤波后得到的估计序列小波分解到各尺度上;然后在各尺度上,利用该尺度上的传感器观测数据对小波分解系数进行更新;最后利用小波重构,达到更新原始估计序列的目的。仿真实验表明,该算法具有很好的数据融合效果。

Multi-Sensor Data Fusion Based on Multi-Scale Kalman Filter

With the analysis on the multi-scale dynamic system model,a novel multi-sensor data fusion algorithm based on wavelet transform and Kalman filter is proposed.The algorithm combines real-time and recursiveness of Kalman filter and multi-scale characteristics of wavelet transform and it is also able to fuse the observed data of multi sensors effectively.Firstly the primal estimate from Kalman filter on the thinnest scale is decomposed by wavelet to each scale.Secondly information data on each scale is refreshed by observed data of corresponding scale using Kalman filter.Finally wavelet reconstruction is applied to integrate the estimate information.After simulation test,the accuracy of data fusion goes well.