小波变换域估计观测噪声方差的Kalman滤波算法 及其在数据融合中的应用

众所周知,卡尔曼滤波的成功应用需要事先准确知道观测噪声的统计特性.本文首先简要分析了不准确的观测噪声统计特性对卡尔曼滤波性能的影响,然后利用小波变换可以实时分离信号和噪声的特性,提出了一种在未知观测噪声条件下的卡尔曼滤波算法,该算法可以实时跟踪观测噪声的变化,即实现了对观测噪声方差的实时估计,从而解决了在未知观测噪声的条件下卡尔曼滤波失效问题.最后讨论了提出的方法在信息融合中的应用,仿真结果证明了本文方法的有效性和实用性.

Kalman Filter with Wavelet-Based Unknown Measurement Noise Estimation and Its Application for Information Fusion

It is well known that the successful applications of the Kalman filter are dependent on whether the prior knowledge of the statistical characteristics of the measurement noise is known. In this paper, the effects of the inaccuracy of the measurement noise covafiance on the filter performance are first analyzed briefly. The feature of the wavelet transform separating a noise signal into the signal and noise parts in real time is combined into Kalman filter. A new method,making the Kalman filter under unknown measurement noise covariance condition valid, is then proposed. The presented method can track the changes of the measurement noise covariance and estimate the covariance in real time. Finally, the applications of the proposed method for the information fusion are discussed. The simulation results verify the effectiveness of the proposed method.