论文部分内容阅读
本文旨在分析偏差校正法与其他参数辩识方法的关系。在前面部分,借助移位向量把二次准则表达成待辩识参数的显函数。这样广义最小二乘法,偏差校正法及伪线性法的离线算法便很容易地以这同一准则导出。后面部分从递推预报误差法导出以上三种方法的递推算法。通过以上分析并采用仿真手段清楚展现了各种算法间的相互关系。
The purpose of this paper is to analyze the relationship between bias correction and other methods of parameter identification. In the previous section, we use the shift vector to express the second criterion as the explicit function of the parameter to be identified. This generalized least-squares method, the deviation correction method, and the off-line algorithm of the pseudo- linear method can easily be derived with this same criterion. The latter part derives the recursive algorithms of the above three methods from the recursive forecast error method. Through the above analysis and the use of simulation tools clearly shows the relationship between the various algorithms.