基于离散小波变换的动态过程模型参数辨识的Markov方法

Markov Method for Parameter Identification of Dynamic Process Model Based on Discrete Wavelet Transform

摘要: 提出一种新的Ｍａｒｋｏｖ递推算法，该算法主要是基于过程噪声的离散小波变换的协方差矩阵及其Ｃｈｏｌｅｓｋｙ分解因子具有特殊稀疏带状结构的事实，利用模型输出误差的小波变换在线估计出其协方差矩阵及其Ｃｈｏｌｅｓｋｙ分解因子Ｌｃｗ，进而利用Ｌｅｗ对数据进行白色化处理。数值仿真验证了所提算法是可行的。

Abstract: Based on fact that the covariance matrix R^ ew of the wavelet transform of process noise and its Cholesky factors L^ ew are specially spasse and band like, the new Markov recursive algorithm can estimate them on line by using the discrete wavelet transform of the model output error and them whiten the input output data. Simulator results have verified the proposed algorithm.