一种子空间辨识对权重矩阵加权平均新方法

摘要

A new method to compute weighting matrices in the linear subspace identification is presented, this method is linear systems based on advantage of three algorithms among N4SID,MOESP and MOESN4,the method MOESN4 is modified and promoted,the main aim to treat the given weighting matrices are used,which makes more choices to the weighting matrices, increases practical application of the presented algorithm. Finally, two kinds of certain different weighted average methods MOESN4WAc and MOESN4WAd are provided.%给出了一种计算线性子空间辨识权重矩阵的新方法，该方法是在结合算法N4SID、MOESP及MOESN4优点的基础上，对算法MOESN4加以改进和推广，主要是对所选取的权重矩阵进行加权平均，使算法的权重矩阵选择机会增多，增加了算法在实际中的应用性，给出两种不同权重矩阵加权平均算法 MOESN4WAc 与MOESN4WAd.