基于Newton-Raphson迭代与PSO数字图像相关方法

鉴于基于Newton-Raphson(N-R)迭代的数字图像相关方法对迭代初值的敏感性问题,提出了一种基于N-R迭代与粒子群优化(PSO)算法的数字图像相关方法。该方法利用了PSO算法中的全局搜索能力与N-R迭代中的局部搜索能力,通过它们的交替迭代以改善初值;以改善后的初值进行最后的N-R迭代。经检验,该方法比N-R迭代对初值的要求宽松。这是由于在N-R迭代与PSO算法多次交替迭代之后,位移和应变的初值都得到了一定的改善。对相似材料模型受载破坏后表面上的点的位移和应变进行了计算,获得了较好的结果。

Digital image correlation method based on Newton-Raphson method and particle swarm optimization algorithm

Since the Digital Image Correlation（DIC） method based on the Newton-Raphson（N-R） method is sensitive to initial values, in the paper, a new DIC method based on the N-R method and the Particle Swarm Optimization （PSO） algorithm is proposed. The new DIC method has the parallel computing capability of the PSO algorithm and local search capability of the N-R method. In searching, the initial values are updated in the process of the alternate iteration of several times by the two methods（N-R and PSO）. The N-R method is used as the last iteration, in which the updated initial values are used. It is found that compared with the N-R method, the new DIC method is not strict to the initial values. This is due to the fact that the initial values of the displacement and strain have been updated since the two methods iterate alternately several times. The displacement and strain of points on a surface of a simulated material model stressed are calculated during the failure and satisfying results are obtained.