摘要
以残差平方和 Q=[y_i-Af(x_i,B,C)-D]~2这一四维函数为目标函数,进行无约束最优化计算方法的研究。通过实例介绍具体计算方法。计算结果表明,本方法对 y=Af(x,B,C)+D 型曲线的拟合,比一般的无约束最优化计算方法,例如麦夸法,有计算简便且易于掌握的优点;比先采用线性化,然后进行最小二乘法的拟合方法,拟合效果更好。
Researches computational method for unconstrained optimization, using sum of residual squares——Q=[y_i-Af(x_i,B,C)-D]~2(four parameters funotion)as objective function.Some examples are used to introduce the method in detail.Results show that the method is simpler than general computational methods for unconstrained optimization e.g.Marquardt algorithm,and fits to four parameters curve better than methods that a curve is linearized prior to giving least-squares fit to the curve.
关键词
非线性
曲线拟合
优选法
方程
参数
nonlinear curve fitting
optimum seeking method
equation
parameter
