一种基于神经网络的卡尔曼滤波改进方法

卡尔曼(Kalman)滤波是一种基于最小方差估计的递推式滤波方法,它要求信号的状态模型是已知的,这就限制了它在实际中的应用.利用神经网络的良好的非线性映射能力对实际系统进行系统辨识,可以获得符合精度要求的系统状态方程,很大程度改进了卡尔曼滤波的效果.相对于一些经典的卡尔曼滤波改进算法,这种方法具有应用范围广和数学建模简单易行的优点.将神经网络与卡尔曼滤波相结合的方法用于图像复原实验,结果表明,该方法具有可行性和有效性.

An Improved Method of Kalman Filter Based on Neural Network

Kalman filter is a recursive filtering method based on minimum variance estimation,but it assumes that the signal's state model is exactly known,which restricts its application in practice.Nevertheless the system's state equation can be obtained through the identification of systems by using neural network's good abilities of non-linear mapping.In contrast to some classic improved algorithm of Kalman filtering,this method has the advantages of wide application range,simple and feasible mathematical modeling.In this paper,the method which integrates neural network and Kalman filter is implemented for image restoration.The experiment result shows that the provided method is effective and available.