摘要
以神经网络作为正向模型,蚁群优化算法作为反演方法,对采集的疲劳裂纹涡流检测(eddy current testing,ECT)信号进行了反演,重构了裂纹形状。研究了算法中参数的不同选择对反演结果的影响。裂纹形状重构结果表明了神经网络正向模型的有效性和蚁群反演算法的可行性。
The reconstruction of crack shapes from eddy current testing （ECT） signals is realized by using artificial neural network based forward model and ant colony algorithm. The parameters of the algorithm are modified and the results are analysed. The results of crack shapes reconstruction validate both the efficiency of the forward model and the feasibility of the inverse approach.
出处
《科学技术与工程》
2008年第6期1545-1549,共5页
Science Technology and Engineering
基金
广东省工业攻关计划(2006B12401001)资助
关键词
自然裂纹
涡流检测
神经网络
正向模型
蚁群算法
形状重构
natural crack eddy current testing artificial neural network forward model ant colony algorithm profile reconstruction
