论文部分内容阅读
本文对连续-离散系统的极大似然估计方法提出了一种灵敏度初值的正交计算法,它是对具有递推计算灵敏度的修正牛顿-拉夫森算法的改进。文中通过借用正交试验的概念和正交表的性质,解决了原有算法的初始矩阵求逆所存在的困难。
In this paper, an orthogonal method of initial sensitivity is proposed for maximum likelihood estimation of continuous-discrete systems, which is an improvement of the modified Newton-Raphson algorithm with recursive computational sensitivity. By borrowing the concept of orthogonal experiment and the property of orthogonal table, this paper solves the difficulties of original matrix inversion of the original algorithm.