【摘要】 在非线性回归模型参数拟合问题中,当数据中的每个变量都存在不可忽略的误差时,在普通的最小二乘准则下拟合出的参数不是最优的.按照总体最小二乘准则,以观测点到拟合曲线或拟合曲面垂直距离平方和为目标函数,然后用最优化方法搜索出使目标函数值取最小值的参数和数据点估计,从而给出求最优模型参数的算法,最后,通过计算机仿真和与文献比较,验证了提出方法的正确性.更多还原

【Abstract】 In nonlinear regression model parameter fitting problem, when each variable of data has negligible error, the fitting parameters by ordinary least squares criterion are not optimal. According to the total least squares criterion, the optimal objective function is sum of vertical distance squares between observation points to the fitting curve or surface fitting,and we use optimization method to find the optimal parameters and concomitant variable estimation values of minimizing objective function. The algorithm is proposed to calculate optimal parameters and concomitant variable estimation. Finally, the correctness of our new method is verified by simulations.更多还原