改进差分进化算法求解B样条曲线曲面拟合问题  

SOLVING B-SPLINE CURVE AND SURFACE FITTING PROBLEMS BY IMPROVED DIFFERENTIAL EVOLUTION ALGORITHM

作　　者：李小艳 陈绍平[1] Li Xiaoyan;Chen Shaoping(School of Science,Wuhan University of Technology,Wuhan 430000,Hubei,China)

机构地区：[1]武汉理工大学理学院,武汉湖北430000

出　　处：《计算机应用与软件》2018年第3期275-281,298,共8页Computer Applications and Software

摘　　要：利用B样条进行数据拟合的关键在于B样条参数(节点矢量和控制顶点)的选取,同时把节点向量和控制顶点视为变量,拟合问题就演变为多维多变量高度非线性的最优化问题。由于差分进化算法(DE)在处理数值优化问题时相比于其他基于种群的进化算法收敛速度更快、稳定性更好,提出一种改进的差分进化算法来处理带噪声数据点的B样条曲线曲面最小二乘拟合。试验产生了多重节点。将其与基本的差分进化算法的试验结果进行比较,得到的BIC值和残差平方和更小。he key to data fitting using B-spline is the selection of B-spline parameters(node vector and control vertex),while the node vector and control vertex are regarded as variables.The fitting problem evolves into a highly nonlinear multi-dimensional multivariable optimization problem.Because the differential evolution algorithm(DE)has faster convergence and better stability than other population-based evolutionary algorithms when dealing with numerical optimization problems,an improved differential evolution algorithm was proposed to deal with B-spline curve and surface least squares fitting.The experiment produced multiple nodes.Compared with the experimental results of the basic differential evolution algorithm,the sum of squares of residuals and BIC values was smaller.

关 键 词：B样条 差分进化算法 最小二乘拟合 随机游走 混沌系统 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]