Numerical Method of an Orthogonal Array Optimization

TLDR: In this article, the damping OAO was introduced into conventional nonlinear regression analysis, and the numerical experiment results indicate the superior application performance of the damp OAAO.

Abstract: The nonlinear regression analysis (RA) is widely used in signal and image processing, but the ill-conditioning problem and the convergence property of classical nonlinear RAs depend on the initial parameter estimates in a great degree. The orthogonal array optimization (OAO) is one of the approaches to get the initial estimates that are near to the optimal values. We introduced the damping idea into conventional OAO, hereby termed as the damping OAO. This paper presents the numerical method of the damping OAO. The numerical experiment results indicate the superior application performance of the damping OAO. It is well known that the nonlinear regression analysis (RA) is widely used in signal and image processing fields. In statistics, nonlinear RA is a form of RA, where observational data are modeled by a function, which is a nonlinear combination of the model parameters and depends on one or more independent variables. Nonlinear least squares method is the common form of nonlinear RAs. The basis of the method is to approximate the model by a linear one and to refine the unknown parameters by successive iterations; and the initial parameter estimates is highly significant for the ill- conditioning problem and the convergence property of the method (1). There are many approaches to get the initial parameter estimates that are near to the optimal values, one of which is the orthogonal array optimization (OAO) (1-2).