一种新的正交参数选优法及其在非线性回归分析中的应用

A NEW ORTHOGONAL OPTIMIZING PARAMETERS METHOD AND THE APPLICATION IN NONLINEAR REGRESSION ANALYSIS

非线性回归分析是工程中经常采用的一种用来估计数学模型参数的方法 ,该方法能否顺利运用与参数初始值的选择有极大关系。本研究提出一种新的正交参数选优法阻尼正交表法 ,它不仅可以保证非线性回归分析算法的顺利收敛 ,而且能够显著提高后者的收敛速度 ,进而极大改善非线性回归分析算法的应用性能。本研究的数值试验及心肌造影超声心动图定量分析应用实例表明 ,作为对传统正交参数选优法的一种改进 。

Nonlinear regression analysis is usually used to estimate the parameters of a mathematical model in engineering. Selecting proper initial values for the parameters is a very key step which determines wether the consequent analysis successful or not. In this paper we present a new orthogonal optimizing parameters method, damp orthogonal array method, which not only makes the algorithm of nonlinear regression analysis convergent, but also accelerates the convergence of the algorithm notably, and consequently, improves the applied property of the algorithm of nonlinear regression analysis greatly. The numerical experiment and the example of applied quantitative analysis to myocardial contrast echocardiography (MCE) indicate that damp orthogonal array method has wide perspective of applications in science and engineering calculation and signal and information processing.