摘要
针对一类特殊的参数辨识问题——椭圆定形曲线拟合问题 ,对原有的几何距离法提出了进化算法的求解方案 .基于反问题的求解思想给出了一种新的椭圆拟合度量 ,和原有度量相比 ,它既反映了拟合曲线的弯曲特性又便于数值求解 ,在该度量基础上构造出一种新型算法——加权新度量法 .数值模拟结果表明 ,提出的新方法是有效的。
Known the basic form of a function,guess the parameters from the observed data, to get its integral expression, which is called a parameter recognition. There have existed some algorithms for solving the ellipse curve fitting problem——a special kind of parameter recognition problem, such as algebra distance method, geometric distance method and weighted least square method, which have their advantages and disadvantages respectively. This paper proposed an evolutionary algorithm (ES) to solve the aforementioned geometric distance method, and provided a new metric to describe the accuracy of ellipse fitting from the thought of resolution of inverse problem. The new metric not only characterizes the curved property of the fitting curve but also leads to an optimization problem convenient to compute. The numerical results show that the new method is effective and feasible.
出处
《上海交通大学学报》
EI
CAS
CSCD
北大核心
2002年第8期1210-1213,共4页
Journal of Shanghai Jiaotong University
